Heart attacks and strokes are big health problems around the world. They are the main reason why people die. Whether someone might get these diseases depends on different things. There are some factors that people can change. These include how much cholesterol is in your blood, blood pressure, how much you exercise, weight, educational qualifications, whether you smoke, whether you are rich or poor, and how much time you spend with friends or family.
There are also factors that people can’t change. This includes age, whether you are male or female, ethnic background, genes, and whether your family has a history of heart disease.
We will create a heart disease risk calculator. This uses information about cholesterol and blood pressure, genetic information and social factors. This tool will help predict how likely someone will get heart disease in the next ten years.
By using information about a person’s genes and social background, we can make this tool more accurate than the ones we have now. Getting an accurate prediction of your risk of heart disease is vital. Knowing the risk of getting heart disease allows you and your healthcare provider to make better choices about living healthier, like changing your diet or starting new medicines.
This research could change the way we treat heart diseases. It might lead to more personal and effective ways to prevent these diseases, helping reduce how often they happen and making people healthier.

The British population is currently the world’s most well-studied in the field of human medical genetics. This 5-year project extends the UK Biobank to a 3rd dimension, by adding DNA from more than 1,000 individuals who lived in Britain during the past 4,500 years. Performing the first fine-scale studies of ancient DNA from a single place will allow us to understand human evolution in response to epidemics, dietary shifts, urbanisation, and industrialisation. It will also aid archaeology and medical genetics by untangling the complex and rich ancestry of present-day British populations from Stone Age, through the Middle Ages to today. Collaborating closely with archaeologists, the project will contextualise DNA information about studied individuals, creating new ways to understand the complex tapestry of ancestry through the human past. By understanding how human biology evolved in the past, we will be better prepared to tackle health challenges today.

Traditional medical imaging, such as X-rays, CT scans, and MRI scans, have been instrumental in diagnosing and treating a wide variety of medical conditions. However those 2D image or 3D volume data are difficult to interpret and visualize in a way that accurately represents the full 3D structure of the body. This can make it challenging for medical professionals to get a complete picture of a patient’s anatomy and identify abnormalities. This inspired lots of research in the past 20 years on medical image segmentation and 3D visualization. Given high resolution imagining and dense scan, the state-of-the-art techniques can reconstruct the 3D model accurately.
However, many medical imaging methods, such as X-rays and CT scans, use ionizing radiation to create images. While the doses used are generally considered safe, repeated exposure can increase the risk of cancer or other radiation-related health problems. Some medical imaging methods can be time-consuming and require patients to lie still for an extended period of time. This can be difficult for some patients, such as those with claustrophobia or mobility issues. Besides, medical imaging methods can be expensive in terms of the equipment and staff cost. This leads to only a very limited imaging could be acquired for each patient, which is normally not enough for the 3D reconstruction to show a complete visualization of the patient’s problem.

Our goal is to reduce the number of medical image captures and supply a 3D visualization of patient’s anatomy using machine learning techniques. Medical data are normally multimodality (image, text, audio, video etc), complex with information from various channels. Machine learning technique have been proven very efficient in processing large data, learn their structures, extract useful features, perform automated analysis, enabling faster and more consistent reasoning. This can improve efficiency in medical workflows, reduce errors, and facilitate more informed decision-making by healthcare providers.
Our project aims to develop a few new machine learning models to 1) train a parametric 3D human skeleton model that can represent individual patient skeletal structure, and 2) efficiently and accurately reconstruct 3D human anatomy from medical images and report.
Our project is expected to last three years and will have a significant impact on productivity in the NHS at scale. By enabling doctors to make more informed decisions and helping professionals and patients understand medical conditions and treatments, our technology has the potential to improve healthcare outcomes.

1) 3D Reconstruction of Head and Trunk Skeletons and Soft Tissue Organs: Using full-body MRI and DEXA data from the UK Biobank, we plan to reconstruct the 3D shapes of the head and trunk skeletons, as well as major soft tissue organs (such as the lungs, liver, heart, spleen, artery, etc.).

2) Segmentation and Reconstruction of Brain Regions and Blood Vessels: Develop AI-based tools to automatically segment and reconstruct brain regions and blood vessels by analyzing structural MRI, MR angiography (MRA), ultrasound imaging (US), and computed tomography angiography (CTA) data. This will aid in identifying key brain areas, optimizing neurosurgical planning, and providing precise navigation support for cerebrovascular surgeries.

3) Generation of Internal Organ and Tissue Shapes Based on RGB Images: By analyzing the correlation between the shape of the skin surface and internal organs and tissues, we will develop a system for generating 3D shapes of internal organs and tissues based on RGB images. This system will utilize deep learning and advanced image processing techniques to infer the shapes of internal organs and tissues from external images.

This project aims to develop new tools to better predict a patient’s risk of genetic diseases. The tool is developed using machine learning techniques and makes a disease risk assessment based on the patient’s genetic data and existing biological knowledge. Besides providing a risk analysis, the project also aims to understand these risk predictions. By interpreting how and based on what genetic factors the prediction for patients is made, we aim to gain insides in the underlying disease mechanisms and which genetic factors are involved.
The project is part of a larger genome interpretation project carried out by the lab of Yves Moreau. The past decades faster and cheaper techniques to read out DNA have created a large amount of genetic data. This opens the possibility to investigate the complex relations between genetics and diseases by using sophisticated and data-hungry machine learning techniques. Ultimately, this will provide diagnostic tools that help clinicians to assess disease risk more precisely and lead to a deeper understanding of how genetic factors are causing diseases, possibly providing new treatment options.

Social relationships give support and meaning in our lives, contributing to good mental health. However, the precise way in which social support leads to good mental health is unclear. There are several interacting factors to consider, such as the type of social support (Hakulinen et al., 2016), socioeconomic conditions (Berkman & Krishna, 2014), and an individual’s personality (Kendler et al., 2002). There is also a dynamic interplay between biological and social factors, as the structure and function of the brain affects receptivity to social support, and social support shapes neural structure and function (Lamblin et al., 2017). Previous research has looked at these different factors individually, however this project aims to build on this to combine biological, psychological and social factors into a model of the mental health benefits of social support.

A statistical method called structural equation modelling will be used to test different relationships between different types of social support on mental health outcomes, the impact of social and personality factors, and how it is related to the structure and function of the brain. The project will use data from the UK Biobank which includes extensive questionnaire data as well as brain imaging data in a very large sample of middle- to older-aged adults. It will form the first study of a PhD and last around 18 months.

The findings of the project could help to create new approaches to improving mental health. By understanding how the brain is affected by social factors and how this relates to anxiety and depression, we can develop new evidence-based approaches to treating and preventing mental health problems that focus on enhanced social support. Loneliness and social isolation is a rising concern, especially in older people, and it is essential to understand this at a biological as well as a social level.

All studied human traits are partly genetic and partly environmentally influenced. While
geneticists have gone to great lengths to find genes that predict human traits, social science has yet to catch up and consider the involvement of genes towards explaining aspects of the human condition. How well do we explain and predict health, education, fertility or mortality based on knowledge about someone’s behaviour (social) environment? How well do we explain and predict human behaviour based on our theories? How much of the social relationships we observe may actually be influenced by genetic causes?

In order to evaluate currently ongoing, costly large-scale data collections and research initiatives, we need a robust assessment to evaluate our reasoning about the social world, the quality of our policy recommendations, and timeless debates about nature versus nurture. The UKBiobank provides us with the unique opportunity to analyse amazingly rich information, whilst considering the interactions among genes and the environment. Central questions that we aim to answer are therefore: To what extent does the (social) environment explain and predict our behaviour or health? How well are we doing in collecting the right data? Do we need more advanced social theories, or do we need to do a better job at collecting data to improve our findings?

In this project, we will, over the course of 36 months, isolate environmental influences from genetic influences to quantify the importance of environmental factors that contribute to various health, status and demographic characteristics of individuals. Furthermore, we aim to evaluate whether future research requires more advanced thinking or more data when it comes to explaining or understanding human conditions. Results may question the stability of our current social science explanations, and highlight the need for complexity in our reasoning about the (social) world. Finally, we approach the complex interplay between genes and the environment and issues of genetic selection in order to approve substantial and statistical knowledge for general knowledge and future research directions.

Aims: To produce a predictive outcome model framework with which to analyze and quantitatively characterize the associations of risk factors and outcomes in an obese population through multivariable regression analysis
Rationale: A risk prediction model of this type does not yet exist for the obese population which generally characterizes the links between risk factors and outcomes. Utilizing a large dataset available within the UK Biobank would serve as a foundation for the creation of such a tool and provide room for expansion and iterative improvement in the future.
Duration: 3 years with annual updates
Impact: Understanding and predicting the progression of obesity would be invaluable towards shaping adverse outcome prevention, care, and disease management. In addition, by basing the project in an ongoing data collection project, there is always the possibility of iterative improvement in the future.

Aim: To investigate whether herpes-zoster and herpes simplex virus-related diseases are associated with frailty
Scientific rationale: Recent reports showed that herpes zoster and herpes simplex viruses-related diseases were associated with neurodegenerative diseases including Alzheimer’s disease and cardiovascular diseases which are major risk factors of frailty in older adults. Herpes zoster and herpes simplex viruses -related disease are frequent in older adults because those people have decline of immunological function, which enable these viruses reactivate and cause related-diseases in those. However, it remains unknown whether those herpes virus-related diseases are associated with frailty.
Project duration: 2024 to 2029
Public health impact: Preventions of those herpes zoster and herpes simplex virus-related disease with vaccination or rapid prescription of medications against these herpes viruses will contribute to prevention of frailty in middle-age and older adults in case that the study show that those herpes virus-related diseases are associated with frailty in those.

Aim:
We aim to provide individualized risk prediction for type 1 diabetes mellitus (T1D) combining both clinical measurements and genetic data.

Scientific rationale:
T1D is a complex disease caused by both genetic and environmental factors such as lifestyles. Genetic factors include single-nucleotide changes that could alter functionality of proteins that could further lead to transition from physiological state to disease state. Many of the nucleotide changes have been discovered. A multitude of these changes have been found to be enriched in T1D patients compared to unaffected population.
Leveraging the rich datasets from UKBB, we aim to develop a prediction algorithm that would calculate individualized risk for T1D by combining effects from the nucleotide changes associated with T1D disease throughout the whole genome.
Additionally, we would investigate the interplay between genetics and clinical presentation including age, sex, glucose level, family history and biomarkers to further understand how genetics interacts with each component to influence disease susceptibility.

Project duration: 09/2021-09/2024
Public health impact:
The goal of developing a prediction model for T1D could potentially identify at-risk population for T1D when they are at pre- or early clinical stage. Our proposed model will add to the clinical tool box for risk prediction for T1D. It may later be applied to clinical consultation, disease management or personalized treatment. Implementation of this predictive scheme would potentially help clinicians to fine-tune the approach of caring or treating T1D patients to mitigate long-term impact on their lives.

Anatomically Modern Humans (AMH) originated and adapted in Africa for a long time before moving Out of Africa(OoA) around 50000-100000 years before present and underwent several demographic transitions in the journey of peopling the Globe. During this journey, AMH faced a different environment than the African subcontinent environment. Over time, AMH subpopulations moved into different parts of the globe and adapted to the specific local environment. This journey draws attention to the constant interaction of AMH populations with the environment i.e. Genotype-Environment(GxE) interaction. The constant GxE interactions resulted in spatially varying phenotypic traits across populations living in a local geographical space. This research work aims to investigate the role of GxE interaction in the local adaptation of human populations. The study aims to develop Machine Learning and Deep Learning model for detecting the genetic variants associated with environmental factors and selected for adaptive phenotypic traits. Identifying genetic and environmental factors responsible for the phenotypic trait variation across populations is crucial in public health and this study sheds light on some of the factors responsible for spatially varying phenotypic traits across human populations residing in a local geographical space.

Research question: The magnitude of nervous system and brain structure changes induced by cardiovascular diseases is unclear.
Aims: We are committed to exploring unknown territory of heart-brain cross-talk, seeking out novel evidence and deciphering the underlying mechanisms. By doing so, we aim to establish new diagnostic approaches and formulate innovative therapeutic strategies that could potentially revolutionize the way we address the complex interplay between heart and brain health.
Scientific rationale: In cardiac dysfunction strong haemodynamics and neuronal signaling feedback interactions between heart and central nervous system exist that are able to bidirectionally provoke acute or chronic functional impairment. Cerebral injury secondary to cardiac dysfunction may include sudden stroke, cognitive decline (which might end up as dementia), and depression. Brain stem functions are related to the heart-brain interaction. In cardiac dysfunction, it’s known that the neurohormonal control and neuronal reflex circuits get out of balance. In this project, we will explore and identify the principal aspects of the pathophysiologic interactions between heart and brain that are associated with cardiac dysfunction. These aspects encompass a range of conditions and mechanisms, including stroke, the impact on cognitive function and brain structure, the relationship with depressive disorders, as well as alterations in neurovegetative control and the regulation of neuronal cardiovascular reflexes.

Genome-wide association study (GWAS) is a statistical tool for detecting associations between genetic variants and phenotypic traits. It is widely applied by scientists studying human, animal and plant genetics. During the past 15 years, many fast GWAS algorithms have been published to cope with the rapidly growing data size. But these algorithms do not necessarily yield consistent results for the same data set due to the implementation of distinct mathematical models and techniques. Thus, we can ask two questions: 1) What are the influence of the mathematical techniques implemented in different algorithms on the result of GWAS? 2) Given a certain set of parameters characterizing the data set (such as the population size, the complexity of genetic architecture), are some algorithms more suitable to apply than others? These questions are important because the results of GWAS suggest candidate regions in the genome which is usually a first step for finding causal genetic variant for important traits, such as complex deceases.
The aim of this project is to provide a knowledge-based solution to the two problems. First, we will provide an in-depth overview of key mathematical techniques implemented in commonly used fast GWAS algorithms. Then, we will select representative algorithms (Some algorithms may essentially implement the same technique despite that they are in different software packages) for the next step, namely a benchmarking analysis. We will compare the results of selected algorithms on empirical data sets across different species (among which human is of course an important one) and evaluate their computational efficiency (for this purpose data sets from UK Biobank is crucial because of its large scale). But since the causal genetic variant is largely unknown in empirical data sets, we will conduct a comprehensive simulation study to evaluate the statistical power and false positive rate of different algorithms. Based on real genomic data from different species, simulated traits with distinct population size, heritability and genetic architecture will be generated according to mathematical models, and each algorithm will be applied to each simulated data set. By combining the results of theoretical review, empirical data analyses and simulation study, we will be able to find out the answer to the two questions mentioned above. We expect to provide a summary on the pros and cons of different techniques implemented in the algorithms as well as recommendations on the choice of algorithms for the research community.

Early onset colorectal cancer (EOCRC) is becoming more common, and it’s essential to understand what causes it. Several factors have been suggested as potential causes, including age, genetics, diet, physical activity, obesity, and more. Nuanced risk factors – like telomere length and body composition – have also been suggested, but are under-investigated. It’s important to comprehensively evaluate all these factors together, and also understand whether and how these risk factors impact EOCRC specifically (by comparing their impact on colorectal cancers in persons >50). This study aims to answer two important questions. First, it wants to understand the various factors that might lead to EOCRC, including telomere length and body composition. Second, we compare risk factors between EOCRC and colorectal cancer that occurs later in life. We will do this by conducting a large analysis of the UK Biobank data, using univariable and multivariable logistic regressions. We will explore a wide variety of risk factors, and evaluate their association with EOCRC and later-onset CRC. The project is expected to be completed and published within 24 months. The public health impact is substantial. EOCRCs have increased by >2% annually, and we need to understand risk factors so we can prevent and screen for EOCRC. We will use the generated results to then risk-stratify persons and ultimately, recommend screening strategies based on personalized risk.

Gastroenteropancreatic neuroendocrine neoplasms (GEP-NENs) are a group of rare cancers that form in hormone-producing cells of the digestive system. The number of people diagnosed with these cancers is increasing worldwide, but we still don’t fully understand what causes them or how to best treat them.

Our research aims to substantially improve knowledge of the biological basis of GEP-NENs by analysing genetic, biochemical, imaging, lifestyle and health data from a large group of people. We will study information from an existing group of 800 GEP-NEN patients as well as from over 500,000 volunteers in the UK Biobank.

By comparing data from people with and without these cancers, we hope to validate known risk factors and discover new biological markers and treatment targets. We will use advanced statistical and computational methods to determine which factors are most important in causing GEP-NENs to develop and progress.

The scale and depth of data in the UK Biobank provides an unprecedented opportunity to make meaningful discoveries about these understudied cancers. Combining findings from our patient cohort with the Biobank data will allow us to develop robust predictive models to improve diagnosis and prognosis. We will also identify the most promising drug targets to accelerate development of new therapies.

Importantly, we will apply rigorous analysis methods to ensure the reliability of the risk factors and biological targets identified. All data and findings will be shared openly with the research community to maximise scientific impact.

This 3-year project will generate high-impact publications and data resources that substantially advance understanding of what drives GEP-NENs. The most significant public health impact will come from discovering new biomarkers for earlier diagnosis and treatment response monitoring, and from identifying validated drug targets to catalyse development of better therapies to improve and extend patients’ lives.

Leveraging the power of the UK Biobank in combination with our unique patient cohort will help transform outcomes for people with these rare but devastating cancers. The rigorous approach ensures the results will have maximal scientific impact and clinical relevance to drive much-needed progress against GEP-NENs.

Sarcopenia, characterized by the age-related loss of muscle mass and function, is closely associated with increased risks of falls, hospitalizations, and mortality in older adults. Early detection of sarcopenia is crucial to mitigate its progression and minimize its impact on health outcomes.
The etiology of sarcopenia is both multifactorial and complex. Numerous factors, including malnutrition, physical inactivity, age-related cellular changes, oxidative stress, inflammation, and hormonal imbalances, contribute to the age-related decline in muscle mass and strength with age. This complexity, along with its association with various other health-related conditions, has driven growing interest in identifying reliable biomarkers to monitor its progression.
To date, most reported studies have primarily focused on identifying single biomarkers for sarcopenia. However, given the “multifactorial” nature of sarcopenia involving multiple biological pathways, relying on a single marker is unlikely to provide sufficient or reliable information. Instead, a multifaceted approach that integrates clinical data, biological insights, and multi-omics analysis is essential for accurately classifying and assessing older adults with sarcopenia.
Despite growing interest, a significant research gap persists, particularly in studies combining clinical data and multi-omics analysis through artificial intelligence (AI). Addressing this gap, our three-year project aims to uncover novel risk factors for sarcopenia by utilizing clinical data, multi-omics insights, and advanced AI methodologies.

Aims:
Cancer is a leading global cause of death. Nearly half of all cancers can be prevented through the avoidance of risk factors and the implementation of prevention strategies. Furthermore, early diagnosis and timely treatment increase the chances of curing many types of cancer. This research aims to identify the factors responsible for genetic changes that lead to cancer and to create early detection methods for various cancer types.

Scientific rationale:
Cancer evolves through stages, starting from normal cells transforming into pre-cancerous lesions and eventually forming malignant tumors. This multi-stage process is influenced by genetic and environmental factors interactions. Identifying the interplay of these genetic and environmental causal factors is crucial for effective cancer prevention. Moreover, reducing cancer mortality is more likely when cases are identified and treated early. Early detection is mainly based on two key approaches: early diagnosis and screening. Early diagnosis relies on identifying early symptoms of different forms of cancer. The early symptoms consist of changes in the characteristics of organs with lesions and adverse impacts on cognitive functions, physical ability, visual acuity, etc. On the other hand, the goal of screening is to pinpoint individuals with specific pre-cancer before symptoms appear. Screening is more resource-intensive than early diagnosis, demanding more precise targeted population identification and biomarker threshold detections. The identification of the cancer causal factors and the development of early cancer detection methods involve a thorough review of existing literature; integrating data from various sources like genomics, biomarkers, environment data, imaging data; using machine learning and statistical methods. The UK Biobank will be the source of the dataset.

Project duration:
We are proposing a project duration of 36 months for our research. The consideration of the timeframe is based on the time needed for processing and preparing the dataset for analysis, conducting statistical analyses, and drafting manuscripts for publication. We believe this duration will allow us to conduct a thorough and comprehensive investigation, ensuring the quality and impact of our research findings.

Public health impact:
The expected impact of the research is its potential to prevent cancers through targeted risk reduction and increase cure rates by enabling earlier intervention. By addressing crucial aspects of cancer control, the research aims to lessen the burden of cancer on individuals and healthcare systems.

Early diagnosis of MDD contributes to a rapid and effective intervention and treatment process and thus is in high demand. Structured clinical interview is currently the most dominant way to diagnose MDD and track the treatment response, where the physician determines the patient’s symptoms by verbally interacting with the patient using standardized assessment tools, which are either clinician-administered or sometimes self-rated. Nevertheless, because the diagnostic process is relatively subjective, dependent on the physician’s expertise and the patient’s cooperation, subjected to human factors, and usually time-consuming, its applicability in the population and its diagnostic effectiveness are not ideal. Precise diagnosis of MDD through biochemical tests remains challenging due to the lack of objective physiological indicators, as well as specific laboratory tests. This indicates the need to develop more effective diagnosis methods based on an in-depth understanding of MDD’s pathophysiology. While no single biomarker exists for MDD diagnosis, there is mounting evidence of multiple dysregulated contributing factors, including inflammatory cytokines, endocrine factors, growth factors, metabolic dysregulation and genetic variations in mood disorders. This knowledge could be used to filter out the associated individual parameters to MDD pathophysiology, understand the causal mechanism and develop biomarker panels that aim to profile a diverse array of hormones, cytokines, inflammatory, metabolic, and genetic markers.
Our aim is to look at all the available biological parameters along with demographic and clinical factors that are associated with MDD risk as well as treatment response. Firstly, we will start with global screening from all the available literature to create a cumulative list of all the significantly associated parameters. Next, we will look for information about these parameters from the patient data retrieved from large biobanks such as UK Biobank. Utilising the information, we will create a prediction model that aims to predict disease risk as well as treatment outcome. Lastly, we will validate the developed model using MDD samples from other populations to assess the predictive accuracy, reproducibility and generalisability of the prediction model.
Developing a predictive model for disease risk and treatment outcomes using a data-driven approach represents a significant advancement. This model could address the existing treatment gap and improve overall response rates in MDD patients. The use of a global population in the meta-analysis and the inclusion of large biobanks enhance the generalizability of the developed model. This global perspective contributes to a more inclusive understanding of MDD across diverse populations.

Our current understanding of spectrum of disease outcomes associated with inherited genetic variation in major genes, including ACD, CDKN2A, CDK4, MITF, POT1, SLC45A2, TERF2IP, and TERT, that result in familial melanoma is limited. We plan to use genotype and whole-exome sequencing data available in the UKBB resource to agnostically determine associations with other (non-melanoma) cancers and non-cancer phenotypes through linkage to cancer registry, hospital, and primary care files. Results will be combined with those arising from parallel analyses in an independent large US data resource to identify top ranking genetically-associated cancers and non-cancer conditions. Findings will be used to inform follow up of melanoma-prone families across the globe.

Mental illness cause the heaviest economic pressures to the society among all types of diseases and the patients suffer enormous pain. But the diagnose of the psychiatric disorders is merely based on the subjective diagnose of psychiatrists. Researchers want to develop an diagnose technique which is more objective and accuracy using MRI which is entirely harmless to the patients. One outstanding technique is deep-learning algorithm which is a kind of artificial intelligence (AI) method. We want to distinguish the patients and normal people using the deep-learning algorithm built on MRI. But the proper parameters of the deep-learning model are largely undetermined, so build a deep-learning model for distinguish mental illness patients is much infeasible. As physiology gender is a much robust characteristic of human, which is also an dichotomous variable. We want to built the deep-learning algorithm for gender classification. And then, We would try to transfer the model on mental illness. The project duration is about 12 mouth including data accumulation, preprocessing, building deep-learning model and testing the model.
The present work aims to solve some fundamental necessities for the clinical diagnosis of mental illness, and it may facilitate the development of treatment for mental illness and reduce the economic pressures of society and the suffering of patients.

Sarcopenia is the progressive loss of skeletal muscle mass, strength and function with aging, leading to disability and frailty.

This project aims to explore three questions in cross-sectional analyses:

1. What is the prevalence of low skeletal muscle mass (fat-free mass) and muscle strength in the population?

2. What is the body composition and grip strength of the population and what lifestyle activities and disease conditions might influence fat free mass and strength?

3. What dietary factors influence low fat free mass and muscle strength? This research aims to improve our understanding of the factors leading to the development of sarcopenia which are poorly understood. The Biobank data offers the opportunity to investigate this issue. By understanding the prevalence, as well as lifestyle, disease and nutritional factors that relate to loss of skeletal muscle mass and strength, we will be better able to prevent and treat sarcopenia. This fulfils the purpose of the UK Biobank which is ?to support a diverse range of research intended to improve the prevention, diagnosis and treatment of illness, and the promotion of health throughout society?. The researchers will use baseline data from Biobank to explore the 3 questions outlined above. We plan to understand the prevalence of low skeletal muscle mass (as fat-free mass) and strength in the population. We also plan to find out what lifestyle, diet and other factors might affect fat free mass and strength. Will use standard statistical analysis methods to do this work. The researchers will be based at the University of East Anglia and Plymouth University, Plymouth. We plan to investigate questions 1 and 2 about low skeletal muscle mass (fat-free mass) muscle strength on the full cohort.

For question 3, relating diet to fat-free mass and muscle strength, we plan to use the subset with dietary information available.

To date, although lots of common genetic variants (Single Nucleotide Polymorphisms, SNP) have been identified from genome-wide association studies (GWAS), the role of rare variants (variants present in less than 1% of the individuals are coined) in genome sequencing technology remains unknown in cancer. In addition, multiple diseases share the similar cause genetic variants or genes, such as chronic obstructive pulmonary disease, emphysema, and lung cancer. Identification of them may help the early detection disease and precision prevention.
We will conduct a cross-trait integration study including multiple diseases on the UK Biobank samples to identify the rare and common genetic variants associated with cancer and cancer related traits. We will explore the key genes in different diseases and identify the share genes. Further, we will use bioinformatics algorithm to benchmark the genes which are important for the diseases.
In summary, this project may help identify the genetic background of multiple cancers and cancer related diseases. Key genetic variants and genes will be screened out. The total period is 36 months.

The aim of our research is to use advanced computer techniques to identify different types of Alzheimer’s disease and vascular dementia based on factors such as frailty, brain health, and genetics. By grouping patients with similar characteristics, we can better understand the unique needs and characteristics of each type of dementia and develop more effective treatment plans. If we can better understand the differences in frailty levels in Alzheimer’s disease and vascular dementia types, it will support the idea that frailty should be considered when diagnosing and treating these conditions. Overall, this year-long research study has the potential to improve our understanding of dementia and improve how it is diagnosed and treated.

Precision medicine is a form of healthcare where disease prevention and treatment is tailored to the individual patient. Besides environmental factors and lifestyle, also genetic variation is taken into account. This proposal will aim to simulate potential effects of pharmacological and lifestyle interventions in an individual person. Much of the knowledge we possess about genomic risk factors comes from statistical measures of association from large-scale population studies. The conceptual and practical disconnect between the populations we study and the individuals we want to treat is a major topic in research. The primary goal of this proposal is to develop a methodology based on machine learning to facilitate precision medicine for CVD patients by connecting population and individual genomic phenomena. We aim to develop a so-called data mining-based workbench, which will allow clinicians to carry out thought experiments about the treatment of individual patients using models of CVD risk derived from population-level studies. This will help clinicians understand how these risk factors might be useful for the diagnosis and treatment of an individual, accelerating the translation of genomic findings into the clinic.
The proposed APM-GDM is based on representation learning which means it can be fed with raw data and automatically extract necessary representation for predictions. An ensemble method or a DL network can provide representations at different levels. In neural networks, for example, the output of each of hidden layers is considered as the representation at that level. The higher layers the data belong, the more abstract representations we get for these data. In different studies, these higher-level representations of raw data prove to be very effective for classification or detection problems.
The project will be conducted for three years and intend to use as many individuals as available to satisfy the need for statistical and machine learning.

Genome-wide association studies (GWAS) are useful in identifying genetic variants associated with human diseases. A polygenic risk score (PRS) uses this the results from GWAS analysis for assigning a risk score for an individual. The genetic risk only one part of the story. Various clinical/biochemical parameters as well as life-style and food habits contribute to the risk of developing the common disease. By integrating the genetic and non-genetic factors we will be able to get the absolute risk score for a disease and one can monitor how the risk for getting a disease increases or decreases by changing the life-style, food habits or through medications. This can be very useful because the genetic risk cannot be altered (in general), however lifestyle modifications ill alter the risk of getting a common disease such as type 2 diabetes or coronary artery disease. This is the next step towards preventive wellness using genetics. This new knowledge will lead to the development of new diagnosis, prevention and hopefully the identification of medicinal drugs, and improve the public health.

The goal of this research is to use deep learning techniques to identify connections between a person’s genetic makeup (genotypes) and their physical characteristics or diseases (phenotypes) using data from the UK Biobank. To do this, our aim is to create a new methodology and to identify specific non-coding genotypes (parts of DNA that do not contain instructions for making proteins) that are related to multiple diseases or physical characteristics. After testing this system with UKBB data, we hope to use this system in a manner that would allow us to analyze data from multiple institutions without exchanging information. The rationale is that traditional methods require data from large number of patients; however having large number of samples in single institutions is not possible, however data is sensitive and therefore cannot be shared across institutions. We plan to test this system using a large number of samples from the UK Biobank. The developed tool will perform phenome-wide association study (PheWAS), which is a powerful way for identifying the relationship between the genetic and a wide range of human diseases and traits. By studying large populations, PheWAS can identify genetic risk factors that may be missed by traditional disease-specific studies. This project will require approximately 3 years. The public health implications of PheWAS studies include a better understanding of the underlying causes of many diseases and the identification of new targets for prevention and treatment. Additionally, PheWAS studies can help to improve the accuracy of diagnosis and the development of more effective treatments. Furthermore, PheWAS results can also help to identify population subgroups that are at a higher risk of certain diseases, which can inform public health interventions and improve health outcomes for those populations. In summary, PheWAS studies have the potential to improve our understanding of the genetic basis of disease, and can inform the development of new treatments and public health interventions.

This project aims to understand more about Cardiovascular Diseases (CVDs) and their connections to other health conditions. Using advanced deep learning technology, we’ll investigate the relationships between CVDs and various traits, such as hypertension, diabetes, and obesity. Our goal is to prioritize these connections and identify potential genetic causes. Additionally, we’ll use a method called Mendelian randomization to figure out if these traits might actually cause CVDs. By comparing our deep learning approach with traditional methods, we hope to discover new biomarkers and crucial genes related to CVDs (including disease risk and prognosis) in the UK Biobank data.

Coronary heart disease is a complex condition influenced by genes, environment, and lifestyle. Previous studies have identified many genetic factors linked to CVDs, but we aim to go further. Our new deep learning method, DSpaLaRefiner, allows us to analyze the data more effectively, identifying patterns and relationships between different traits. Unlike traditional methods, our approach does not rely on external data, giving us a unique advantage in discovering new insights. By understanding the genetic correlations and disease susceptibilities and prognosis, we hope to find more accurate ways to predict, prevent, and manage CVDs.

We will use the GATK software for quality control and Plink1.9, ‘survival’ R package, and custom R codes for genome-wide association analysis on the UK Biobank data. DSpaLaRefiner, our in-house deep learning algorithm, will play a crucial role in identifying disease-associated variants and understanding the relationships between different traits. Mendelian randomization analysis will help us investigate if there are causal links between CVDs and other health conditions. This comprehensive approach aims to uncover new insights into CVDs and their connections, potentially revolutionizing our understanding of this complex disease.

The duration of this project is expected to be 3 years; we will consider extending the research duration based on the progress or discoveries made during the project.

High resolution (HR) Cardiac magnetic resonance images (CMRI) provide more anatomical details and enable more precise analyses, and are therefore highly desired in clinical and research applications. However, acquiring such data with an adequate resolution is time consuming. A common way to partly achieve this goal is to acquire MR images with good in-plane resolution and poor through-plane resolution to save scanning time.The aims of the research project is to improve through-plane resolution when saving MR scanning time.Using super-resolution method based on deep learning could use the mapping between the high in-plane resolution images and simulated lower resolution images, to estimate high resolution through-plane images.This project duration will last 24 months.The expected value of the research would produce high through-plane resolution as well as saving Cardiac MRI scanning time.

The main aim is to create a deep learning model based on retinal pictures that can distinguish between people with Alzheimer’s disease and people without dementia with a level of accuracy that is consistent across time.
The most widespread type of dementia is Alzheimer’s disease (AD) and It is a brain condition that gradually robs people of their memory, thinking abilities, and ultimately their capacity to complete even the most basic tasks. Alzheimer’s disease (AD) is a complex neurodegenerative condition with a number of known and improbable causes, as well as a wide range of anatomical features. The retina, which is a part of the central nervous system (CNS), has been called a “window to the brain” and a brand-new indicator of Alzheimer’s disease (AD). Retina tests are appropriate for large-scale population screening and research into preclinical AD because of their low cost, simplicity of access, and non-invasive properties. Additionally, a number of cutting-edge techniques for retinal imaging, such as optical coherence tomography (OCT), have been developed that enable the visualisation of retinal alterations at a very fine resolution.

Therefore, the primary goal is to create a deep learning model based on retinal images to identify people who have Alzheimer’s disease and to consistently perform accurately in differentiating between people who have Alzheimer’s disease and people who do not have dementia with respect to both accuracy and sensitivity.

The Project can be completed within 36 months.

So the public can perform the Alzheimers screening at very low expense or even at free of cost and within short while. At present there is no such facility is available for this purpose in most of the developing countries like India. Right now the public have pay a lot of money for Alzheimers screening (for PET Scan or Brain imaging etc).

Our research project – “A Deep Learning model for the early detection of cardiac amyloidosis among patients with Left Ventricular Hypertrophy”, aims to develop a deep learning based algorithm for classification of various diseases that cause Left Ventricular Hypertrophy – with an emphasis on cardiac amyloidosis among those diseases.

Left Ventricular Hypertrophy is a state in which the heart muscle (myocardium) is over-thickened, and as a result it’s filling capacity is reduced.
The above described state can arise from a number of diseases, one of them is cardiac amyloidosis.
Cardiac amyloidosis is an underdiagnosed, potentially reversible disease, that cause ‘Restrictive Cardiomyopathy’ which leads to progressive Heart Failure with life threatening arrhythmias. The disease arise from abnormal deposition of proteins in a number of tissues, including the heart.

As for today, diagnosis of cardiac amyloidosis is given after a series of tests, including imaging studies, blood tests, scintigraphy and heart biopsy. All of the above makes the journey of the patient towards diagnosis long, expensive and filled with uncertainty.
Given the underdiagnosis and significant morbidity of cardiac amyloidosis, combined with the availability of treatment with disease-modyfing agents, highlights the importance of early diagnosis and treatment.

Several studies has been conducted in order to built a prediction tool for cardiac amyloidosis. Most of them were based on clinical data alone.
We strongly believe that by applying computational learning methods, We’ll be able to built a robust classification tool, that we’ll allow us to predict cardiac amyloidosis as early as possible, and to start disease modifying treatment before the development of progressive heart failure.

In order to do so, we wish to conduct a retrospective study on UK Biobank patients diagnosed with Left Ventricular Hypertrophy based on echocardiographic/electrogram criteria, in order to generate a prediction model for classification of cardiac amyloidosis. Such classification model will enable us to diagnose the disease at an early state, and provide disease-modyfing medication for slowing its progression.
The requested data is a set of several clinical evaluation tools for the patient with heart disease – cardiac MRI (Magnetic Resonance Imaging), Echocardiography, Electrogram and clinical data.
Apllying image processing computational methods on the above will allow us to extract meaningful features for prediction. Combining those with clinical data, will enable us to train state of the art Artificial Intelligence models based on neural networks, that hopefully will be able to classify correctly and thus predict, cardiac amyloidosis in patients with Left Ventricular Hypertrophy.

Our research aims to develop an advanced disease prediction model for complex diseases by integrating both common and rare genetic variants identified through whole-exome sequencing (WES) data. One of the primary objectives of our study is to evaluate the heritability of complex diseases using WES data, shedding light on the underlying genetic factors contributing to disease susceptibility. Additionally, we strive to enhance the prediction accuracy of existing models based solely on polygenic risk scores (PRS). By incorporating rare variants into the model, we hope to capture the additional sources of genetic variation that are often overlooked when focusing solely on common variants.

Previous large-scale population studies have shown that PRS, which consider the effects of common genetic variants, can effectively identify individuals at high risk for complex diseases. However, it is important to note that the overall prediction accuracy of PRS models varies, with most falling within the AUC range of 0.6-0.75. This discrepancy may arise from the fact that most current PRS methods solely rely on common variants and do not fully consider the impact of rare variants on disease susceptibility. Despite their lower frequency, rare variants can possess a substantial effect size and contribute significantly to the development of common diseases, even in individuals with low PRS. By integrating both common and rare variants into our prediction model, we aim to provide a more comprehensive understanding of genetic predisposition.

Our proposed research will leverage the extensive whole-exome sequencing data available from the UK Biobank to identify genetic susceptibility loci associated with complex diseases and develop a prediction model based on WES data. This model will hold great potential to advance our knowledge of genetic risk factors, empowering medical professionals to make informed decisions regarding personalized prevention strategies, clinical interventions, and disease management. By considering both common and rare variants, we hope to gain deeper insights into disease risk prediction and contribute to the broader field of personalized medicine. Ultimately, our research endeavors to reduce disease burden, enhance public health outcomes, and drive evidence-based healthcare interventions tailored to individual genetic profiles.

project duration: 36 months.

Diseases could be viewed as specific sets of phenotypes affecting one or several physiological systems. It has been shown that diseases whose components link to each other at the cellular level show higher comorbidity in the population. Systematically studying entire sets of comorbidities offer a complementary perspective of disease biology from traditional approaches. The overall aim of this project is to systematically study the associations between several diseases in the human population by building phenotypic disease networks (PDNs), with a focus on the known, strong comorbidities between 1) cardiovascular disease and psychiatric disorder, 2) cancer and psychiatric disorder, and 3) rare but complex diseases such as neurodegenerative diseases and psychiatric disorder. Importantly, we will contrast PDNs estimated in the UK Biobank with PDNs estimated using the entire Swedish population in our parallel study using the Swedish national registers.

Exploring comorbidities from a network perspective could help determine whether differences in the comorbidity patterns indicate differences in genetic background, biological processes, environmental factors, or health care quality provided for each population. This project together with our parallel study using the entire Swedish population will offer unique insights into the structure and mechanisms underlying comorbidities in different human populations and will potentially promote the prevention, diagnosis and treatment of diseases and the overall health throughout the society.

Rationale
Dementia is a progressive neurodegenerative disease in which patients encounter frequent delays in diagnosis, leading to increased morbidity. There is a major need of biomarkers for the early prediction as acknowledged by the Alzheimer’s Drug Discovery Foundation. The eye is often seen as the ‘window to the brain’ with much effort dedicated recently to predicting disease such as mild cognitive impairment (MCI) and dementia through images of the retinal microcirculation. The eye is closely coupled to the brain through its circulation, with the eye being perfused through a branch off the internal carotid artery that then continues on to the brain. Furthermore, the eye is coupled to the brain neuronally through projection of the CNS into the retina (neuroretina). As such, it is believed that retinal imaging can be used to monitor changes in the brain which has been demonstrated by recent published data. However, whilst the eye is easily imageable down to the micro-scale, the skull around the brain makes imaging much more difficult. Computational models, on the other hand, can simulate the brain circulation and pressure so we know exactly what is happening in the brain in silico. Whilst utilising artificial intelligence (AI) models retinal imaging can help in the diagnosis and prediction of MCI and dementia.

Aim: To develop a model of the circulation in the brain coupled to the eye, observing retinal circulatory alterations with concomitant change in the brain and development of AI models for diagnosis (detection) and prediction of future disease.

Utilising data in the UK Biobank, his work will entail:
1. Coupling circulation models of the brain and eye (In Silico/digital twin model)
2. Synthetic imaging of the eye and brain pipeline simulation (In Silico/digital twin model)
3. Development of AI model of MCI/dementia detection and prediction of dementia and other diseases

This project aims to investigate how large-scale multimodal data, including fMRI, DTI, T1w/T2w MRI, fundus photography, OCT, and genetic profiles, can be integrated to reveal common biomarkers underlying both neurological and ocular diseases. It further explores whether a foundation model pre-trained on these data modalities can reliably detect early pathological changes in diseases such as Alzheimer’s, Parkinson’s, strokes, cognitive decline, and retinal degenerations, and designs interpretability strategies to elucidate the pathogenic mechanisms shared between brain and eye, potentially informing new therapeutic targets.

The primary objective is to develop a unified deep learning foundation model capable of processing multimodal brain and eye data, leveraging large-scale UKB data for self-supervised pretraining. The model will be evaluated for its performance in early-stage detection and more accurate diagnosis of neurological and ocular diseases, including Alzheimer’s, Parkinson’s, dementia, stroke, schizophrenia, glaucoma, and hereditary retinal disorders, as compared to single-modality baselines. Additionally, the research aims to incorporate explainable AI tools to identify and interpret the critical imaging and genomic features driving disease risk and progression, and to investigate gene-environment interactions by integrating genetic data with imaging for biomarker and therapeutic target discovery.

The scientific rationale lies in the shared developmental origins of retinal and cerebral tissues, which position ocular imaging as a non-invasive proxy for brain health. Recent studies have demonstrated strong correlations between retinal changes and structural or functional brain abnormalities, yet current approaches rarely exploit the full potential of multimodal data integration. The use of interpretable AI will further enable the linking of model predictions with underlying pathophysiology, guiding early interventions and advancing precision medicine.

Complex diseases result from a combination of genetic, environmental, and lifestyle factors. The vast majority of non-communicable diseases are also referred to as complex diseases, including cardiovascular diseases, cancers, diabetes, Alzheimer’s disease, asthma, and many more, and they are responsible for most of the burden on the health care system. Environmental factors acting alone or in concert with genetic or other host susceptibility factors have long been implicated as major contributors to disease burden. Identifying and understanding factors that influence complex diseases is important to guide health-related choices and medical interventions. Yet identification of specific genetic or environmental factors, their interactions, and their effects on human complex disease has remained elusive. An environment-wide association study (EWAS) is a type of epidemiological study analogous to the genome-wide association study (GWAS). The EWAS can systematically examines the association between a complex disease and multiple individual environmental factors. This research aims to search for genetic and environmental factors associated with complex diseases on a broad scale by performing GWAS and EWAS. We also plan to develop a framework for quantifying the contribution of genes, environmental factors, and gene-environment interactions to complex diseases, which can help understand the role of genetics and environment in determining the course of a disease. This project will enhance our understanding of the genetic architecture and pathways of complex diseases. These results generated from this project could inform public health authorities and the general public themselves which risk factors associated with complex diseases should be paid more attention to and the extent to which the changes of environmental and lifestyle factors may help offset the genetic risk of complex disease. This project is excepted to take around 36 months to complete.

In a real-world setting, we have been limited in our ability to infer whether a particular lifestyle or a treatment received by an individual leads to a good or a bad health outcome, despite systems to measure both these lifestyle features and treatments, and the outcomes. A major limitation is that inferring such a cause-and-effect association requires that at least identical individuals exist, which differ only on the lifestyle or treatment in question. The ability to define individuals that resemble each other has, however, been limited by our ability to actually leverage only crude characteristics even though electronic health records as well as electrocardiographic and imaging data may capture some of these distinct patient features.

The current proposal investigates a novel strategy to create an efficient way to define each individual using their data captured in all these high-quality data sources. This virtual representation of each person (referred to as their “digital twin”) will be used to identify other individuals who resemble this person on a set of measurable characteristics.

Our work will evaluate the least amount of unique data sources that are required to define digital twins. Pairs of digital twins will be followed over time for the development of adverse cardiovascular events while focusing on identifying lifestyle or treatment differences that may underlie differences in the trajectory of outcomes among such digital twins.

Collectively, our investigations will leverage the uniquely powerful contributions of UK Biobank participants to make methodologic advances that allow us to gain deeper insights from observational studies, potentially expanding their role in scientific discovery.

The rising of the post-genomic era opened new perspectives for treatment opportunities for many diseases, by allowing researchers and clinicians to focus on the molecular bases of such diseases in an easier fashion. Unfortunately, for most of the genetic disorders, this process is hampered by the limited cohort of patients that limits the knowledge about the driver molecular functions altered by such mutations. This, in turn, leads to an inadequate treatment of patients and to an inefficient healthcare system. In the last years, many researchers tried to build comprehensive tools that have the advantage to lead to a high number of information about the mechanisms of diseases. On the other side, it is very difficult to find a good balance between the cost of the screening and the amount of generated data. In our laboratory, we developed a method to efficiently screen thousands of protein variants in a single multiplexed experiment and with a moderate cost. As a proof of principle, we applied it to P63, a transcription factor that plays a key role in skin development and whose mutations are causative of genetic disorders. To validate the method we aim to identify patients with novel variants that we identified with our screening and “sane” patients that carry potentially pathogenic mutations. To solve this task, UKBiobank will be fundamental, by allowing us to access the genetic data of thousands of individuals with no hard disease-related phenotypes. If confirmed, our screening method could represent a milestone in the field of genetic disorder, by providing the clinicians a key weapon to fasten a specific diagnosis for a disease and, in turn, to adequately take care of the patient with a specific treatment

At present, there are already sufficient studies on peripheral arterial disease, but few studies have focused on whether there are pathogenic gene mutations or abnormal protein function in intracranial arteriosclerosis. Therefore, this study hopes to conduct co-localization analysis of genomics and proteomics, use Mendelian randomization methods for causal inference, and use imaging data for fusion analysis in the hope of discovering the characteristics that lead to intracranial arteriosclerosis and premature intracranial arteriosclerosis.

Intracranial Atherosclerosis (41202, ICD Code: I67.2) is a major contributor to ischemic stroke and cognitive decline, especially in Asian populations. Despite its clinical significance, current research predominantly focuses on extracranial or peripheral arterial disease, leaving the pathogenesis of intracranial arteriosclerosis insufficiently understood. Unlike peripheral atherosclerosis, intracranial arteriosclerosis presents distinct biological behavior and risk profiles, suggesting that different molecular mechanisms may be involved.

Recent advances in genomics and radiomics have enabled multi-layered exploration of complex diseases. However, these modalities are rarely integrated in a unified analytic framework to investigate intracranial vascular pathology. In particular, the genetic architecture underlying intracranial arteriosclerosis remains poorly defined, and it is unclear whether certain protein-level changes mediate the effects of inherited genetic variants. Furthermore, the relationship between these molecular signatures and radiological phenotypes of intracranial artery disease has not been systematically studied.

The goal of this research is to develop a generative model for predicting age of healthy subjects based on their T1-weighted brain scans, in particular from grey matter segmentation maps. The predictive method entails a forward model that uses the target variable (age) and other demographic variables to predict the grey matter image of
a subject. In a subsequent step, the model needs to be “inverted”, to obtain the prediction of age, based on the subject’s grey matter segmentation map and some demographic variables. T1-weighted scans of healthy subjects from the UK Biobank will be processed to obtain grey matter segmentation maps, and then used to train and test the predictive method. The prediction performances achieved by the proposed method will be compared with the ones of current state-of-the-art methods for brain age prediction. The age prediction task is considered as a way to test the proposed predictive method, which can be used later on for prediction tasks in a clinical setting, such as prediction of some clinical scores in Multiple Sclerosis. Furthermore, brain age prediction is of clinical interest itself: several studies have shown that the difference between real age and predicted brain age correlates with some measures of disability. Therefore this “brain age gap” can be used as a biomarker, to study healthy ageing and to characterise pathological deviations underlying several diseases. The expected duration of the project is 2 years.

Human brain changes with normal aging and incidence of brain diseases also rise with age. In order to provide aging people with medical treatments at the right timing, expert routine inspection of brain structure and function in medical images, which is an important proxy to examine brain changes with increasing age, is important for determining if the changes are normal or not. However, this expert inspection can be very time-consuming and expensive. There have been researchers’ efforts to automate this manual inspection of brain MR images using computer algorithms, but it has been hampered by high computing demand and limited amount of brain MR image datasets which are crucial for developing accurate automated computer algorithms. Due to the recent advent of super-computers and big public brain MRI datasets such as the UK Biobank data, developing artificial intelligence (AI) based automated tools for examining and creating brain MR images is being rapidly accelerated.
In this study, we propose developing computer algorithms that can automatically generate a healthy brain medical image of a subject, in which its brain appearance is normal for the subject’s specific age. The computer algorithm will automatically create a normal-for-age healthy brain medical image when the subject’s demographic and clinical information and real brain medical image are given to the computer algorithm. Then the difference between the AI-generated healthy brain image and the real brain image will be automatically analyzed to help physicians decide if the subject’s brain condition is healthy or abnormal for the specific age. This appearance gap between real brain MRI image and computer-generated healthy brain MRI image will serve as a medical feature indicating potential risk of having brain diseases and the severity of them. The new AI based computer algorithm would enable to discover new medical indicators that signify potential risk of a subject to develop further aging-related brain diseases.
We will validate the new computer algorithm with both healthy subjects and patients with brain disease to determine if there is any meaningful relation between the brain appearance gap and medical symptoms in patients with brain diseases. The outcome of the study will help improving the accuracy of routine brain MR exams and reducing the associated expert time and cost, which would be beneficial for managing aging population more effectively. The project will be conducted over multiple years as the UK Biobank dataset continues including more new subjects.

The aim of the proposed research is to understand pleiotropy: that is the nature, extent, and effect of genetic variation on multiple phenotypes. The extent of pleiotropy in humans is an important open question. Apart from inherent interest, it is directly relevant to the use of human genetics for improving drug development pipelines (see below). By their nature, studies of pleiotropy require data on multiple phenotypes. We aim to investigate pleiotropy using genetic data together with baseline measurements and the biomarker data (as it becomes available). Our analysis ultimately aims to inform decisions about which genes and pathways are the best targets for drug development. The efficacy and safety of therapeutics depends on the consequences of perturbations, by the drug, of particular gene products. Genetic variants also perturb the nature or amount of gene products, and is informative for drug efficacy, with effects on other phenotypes informative for on-target safety effects. The proposed work, mainly on non-clinical phenotypes, will involve proof-of-principle studies and development of statistical methods. We will make a further application when more clinical phenotypes are available in UK Biobank The research will use computers to build statistical models of the correlation between the genetic variation in an individual?s genome and biological measurements collected by UK Biobank. We can use the research to ask: if the genetic difference in a gene mimics or is related to the effects of a treatment what is likely to be the (positive and negative) effects of giving it to patients? To do this effectively we will look at the relationship between genetic variation and multiple phenotypes at the same time. Full cohort.

Obesity and fatty liver disease are the results of abnormal fat deposition, which means fat tissue accumulated in an excessive amount in a particular body part. They could increase the risk of life-threatening conditions, such as heart attack. In this project, we aimed to add novel knowledge of the genes and environmental factors behind abnormal fat deposition to the existing literature. We will first calculate the fat content in the whole liver using a software developed by our team. After integrating with other fat deposition phenotypes in the UK Biobank, we will try to find the genes associated with fat deposition, as well as the environmental factors that interact with these genes. The relationship between abnormal fat deposition and human diseases or traits will also be investigated.
With our results, the prevalence of fatty liver disease in the UK could be estimated in a much larger cohort. Our results would also shed light on the genetic mechanism of the diseases related to abnormal fat deposition.
The project is expected to last for two years.

The aim of the current study is to gain further understanding of the genetic factors that underlie human cognition and cognitive decline by identifying the genes that contribute to variation in cognitive function in the UK Biobank.
The project will correlate genetic and phenotypic variation to identify genes that contribute to memory and processing speed (pairs matching, prospective, numeric, light pattern memory, reaction time and fluid intelligence), and develop predictors of cognitive impairment based on genetic markers.
Cognitive impairment is a major health and social issue in ageing populations. Age-related cognitive decline is costly to the individual, his relatives and society in general. It represents a major financial burden to the health services and often precedes dementia, illness or death.
Individual variation in cognitive ageing is partly genetic and partly environmental. About 50% of the cognitive variation is genetic. Identifying the genes that contribute to cognitive ageing would allow developing better prediction models of cognitive impairment, thereby facilitating early intervention; and better understanding of the molecular basis of disease that could eventually provide better or new treatments. Cognitive measurements and their change will be compared to the genetic variations measured from blood DNA. We expect that variation in DNA within, or nearby, genes relevant to cognitive function will correlate with differences in cognitive function among individuals.
We will use statistical methods to test if any of the hundreds of thousands of single nucleotide polymorphisms (SNPs), a type of genetic variation, measured in the UK Biobank is associated with changes in cognition. Subset of the cohort with cognitive measurements that had cognitive measurements at recruitment.

The US Million Veteran Program (MVP) partners with Veterans to study how genes affect health, by safely collecting blood samples and health information from up to one million Veteran volunteers. So far, four beta projects were approved and funded to study Cardiovascular risk factors, Multi-substance use, pharmacogenomics of kidney disease, and Metabolic conditions. Here, we aim to use the genotype and phenotype data from UK biobank to replicate novel findings from MVP, to become a genetic study with unprecedented power for discovery and replication. The goal of MVP is to better understand how genes affect health and illness in order to improve health care for Veterans. This is well in line with the UK Biobank?s aim to improve the prevention, diagnosis and treatment of a wide range of serious and life-threatening illnesses. The research participants of MVP are US Veterans volunteers, their motivation is to help transform health care, not only for themselves, but for future generations of veterans and the general population. We propose to: (1). Use the MVP data as a discovery for genome-wide and pheno-wide analysis to discover novel associations with the four group of traits mentioned above (with a plan to extend to more traits including cancer and neurodegenerative traits); (2). Follow up the novel findings in the UK Biobank data, where we use the same protocol for statistical analyses; (3). Explore the combination of two largest cohorts (MVP in US and UK Biobank in UK) to further characterize the genetic architecture of complex traits with an unprecedented statistical power. MVP will be the single largest cohort with a rich collection of clinical records over two decades. Also, the MVP used the same Affymetrix Axiom genotype array as UK Biobank, which made synergizing these two studies a natural choice. We are requesting to use the full cohort of UK Biobank.

1. Aims: Disease susceptibility and the contributions of daytime napping to these diseases change along with aging. We aims to capture the trend in which daytime napping affects human health conditions in different age groups.
2. Scientific rationale: The role of daytime napping in maintaining body health was said to be different in children and adults. At the same time, academic investigators did not get a putative perspective whether daytime napping is good or bad for people. To solve the puzzle, we designed a study using UK Biobank genotype and phenotype data stratified by age groups. In this way, we are able to infer the casual relationship between daytime napping and many other human traits, hence, bring up a more comprehensive perspective of the consequences of daytime napping.
3. Project duration: Two years will be needed to complete this project.
4. Public health: A proper guidance of daytime napping (e.g. frequency and duration) depending on age will be proposed.

Occupational mental health problems are dramatically increasing during the past few years. It has been very costly to both individuals and the society. Although the past research in social science has explored the varied social and psychological factors that may influence individual’s occupational mental health, the specific genetic variants and neurological foundation responsible for this have largely remained elusive. Therefore, to better understand its biological mechanism and provide with possible career development advice for millions of employees, our research aims to identify the relationship between individual’s genes, brains, and mental health across different occupations. Specifically, we will firstly explore the genome-wide association with employees’ mental health in different occupations, and then further explore the functioning of brains in this relationship. Moreover, we will also investigate the gene × environment interactions on the occupational mental health. We are expected to investigate these research questions in the following three years.

As we age, we accumulate health problems like medical conditions and/or disabilities. However, the speed at which we accumulate those problems is highly variable. On one hand, some people have already accumulated a high number of health problems at a young age, and they look ‘frail’ and/or ‘older than their age’. On the other hand, some people have an advanced age but live with very few health problems, are ‘resilient’ and seem ‘younger than their age’.

The reasons behind this wide variation between age (chronological age) and health problems (biological age) are not well known but there may be genetic reasons. We know that variations in genes that deal with inflammatory reactions in the body have been associated with predisposition to accumulate health problems. However, we also need to consider the environment where those inflammatory reactions take place, which is the so-called ‘connective tissue’. The connective tissue is made of proteins such as collagen and elastin. Collagens are the most abundant proteins in humans, and they provide strength and shape to the tissues. In addition, a lot of chemical reactions (including inflammatory reactions) happen on their surface, and many other cells depend on connective tissue for their normal functions. In addition, collagens and elastin form the ‘building blocks’ of blood vessels, and it is possible that genetic variations in those could have whole-system consequences in terms of resilience or a tendency to accelerated ageing.

We can measure accumulation of health problems using indices of multimorbidity (that is, the number of diagnosed medical conditions) and disability (that is, problems performing activities of daily living); in addition, in UK Biobank there are also validated measures of frailty (such as the frailty index or the frailty phenotype). We propose a Genome-Wide Association Study of these biological age measures in UK Biobank in order to gain a better understanding of the genetic determinants of biological ageing in humans. Understanding the genetic determinants of frailty and resilience in middle-aged and older people could have a significant public health impact.

Our proposed 3-year project will provide an initial phase for a collaboration between the applicant (Prof. Roman Romero-Ortuno: RRO) and Prof. Ross McManus (Professor of Molecular Medicine at Trinity College Dublin). RRO has access to The Irish Longitudinal Study on Ageing (TILDA, http://tilda.tcd.ie) and the proposed pilot work in UK Biobank will inform further collaborative TILDA research looking at novel markers of resilience and successful ageing.

Vasculitis is a rare disease where the immune system -which normally fights infection- mistakenly attacks the blood vessels and surrounding tissues, leading to dysfunction of potentially any organ and system in the body. ANCA-associated vasculitis (AAV) is a type of vasculitis characterized by the abnormal production of antibodies against inflammatory blood cells called neutrophils. These antibodies, known as ANCAs (Anti-Neutrophil Cytoplasmic Antibodies), are directed against one of two neutrophil components, myloperoxidase (MPO) and proteinase-3 (PR3), and activate neutrophils upon binding, leading to inflammation of small and medium-sized vessels.
Disease manifestations can be very different from one patient to the other, ranging from life-threatening (especially when the kidneys, the lungs or the heart are involved) to a relatively benign presentation. Drugs that suppress the immune response are the mainstay of treatment and are usually effective, but can have severe side effects. What causes the disease, and why it can be so different from one patient to the other, is not well understood.
We previously conducted the first large-scale genetic association study in more than 1200 patients with AAV, and identified genetic variants that differentially predispose to the 2 main subsets of the disease, namely PR3-positive and MPO-positive AAV.
This project aims to further look into the genetic architecture of AAV by studying a new cohort of over 3000 patients, recruited from all over Europe thanks to the European Vasculitis Genetics Consortium (EVGC). The genetic makeup of AAV patients will be compared with healthy controls from the UK BioBank, and the results will be combined with those of our previous study. By substantially increasing the number of patients and healthy controls to study, we will have significantly higher chances to discover new genetic variants that affect predisposition to AAV. Moreover, we will look into whether different genetic profiles can explain clinically relevant differences between patients (for instance, whether the vasculitis affect the kidneys, or the lungs).
A better understanding of the genetic architecture of AAV can provide novel insights into biological processes that play a key role in the disease. This line of research can identify promising drug targets, aiding the development of new medicines with an improved safety profile. Moreover, the discovery of genetic markers that modulate disease course may help doctors to better tailor treatment to patients’ needs, improving disease outcomes and quality of life.

Gout is a form of arthritis with the primary cause elevated levels of uric acid. In some but not all people the uric acid crystallises in the joints with gout resulting from a painful immune system response. Genetic causes of elevated uric acid are relatively well understood, however knowledge of the genetic factors controlling crystallisation of uric acid and subsequent immune response are extremely poorly understood. Therefore the aim of the proposed research is to identify genetic variants that influence the risk of gout, particularly those controlling crystallisation and immune response. UK Biobank has the aim of improving the prevention, diagnosis and treatment of a wide range of serious and life-threatening illnesses. Gout is a serious illness. It affects up to 3% of adults in Westernised countries and can be a disabling form of arthritis in some people. It is also associated with other serious metabolic conditions such as diabetes and heart and kidney disease. With our aim to find out why some people get gout and others don’t, the proposed purpose directly meets the UK Biobank’s stated purpose. We will create a group of people with gout and a group without gout, matched by age and sex. Gout will be defined as those with doctor diagnosed gout, as determined from the medical records. We will then take the genome-wide genotype data and scan the genome for genetic variants that have a statistically significant difference between people with and without gout. These genetic variants will pinpoint genes involved in gout. To determine which genes are involved in crystallisation and immune response we will also compare to people with elevated uric acid but without gout. All people of European Caucasian ancestry with doctor-diagnosed gout from medical records (10-15,000). Two age- and sex-matched controls for every case (20-30,000). A separate group of asymptomatic hyperuricaemic controls (20-30,000). We would select the participants, therefore we request access to the entire dataset.

Large-artery intracranial atherosclerosis (ICAs) is a major cause of stroke worldwide. Different clinical risk factors have been associated with risk of ICAs. However, there is still a gap in the knowledge of the physiopathology of the disease. Furtheremore, there is not any drug to prevent or treat stroke in patients with ICAs.

Genome-Wide Association studies (GWAs) have identified 6 loci associated with large-artery stroke. However, no GWAs has been performed to find genetic variants associated with ICAs. Given the relevance that genetics seems to have in this pathology, our aim is to identify genetic risk factors associated with ICAs using a GWAs approach. This data could help to identify key factors for this disease. These factors could be then targetted by existing or new drugs. Some studies from Astrazeneca has demonstrated that drugs developed using GWAs data have a very higher probability to be used in clinics than drugs developed from animal studies.

We have analyzed 107 patients with ICAs (94.6%symptomatic) and 316 population-based controls and we have replicated the results in two cohorts of 444 and 1,270 stroke patients attributed to intracranial and extracranial atherosclerosis, respectively, and 25,643 controls. We have identified one locus at 18q22 associated with the presence of ICAs. This region was replicated in an independent cohort f (p-value=0.009 for the top SNP1). There was no evidence for association of this polymorphism with extracranial atherosclerosis.
This locus is located near a prostaglandin-reductase gene previously associated with VEGF levels. This protein has been previously related with developement and progression of angiogenesis. Angiogenesis is a kew factor in the progression of patients after suffering a stroke. Further studies are needed to really see the implication of the polymorphisms in angiogenesis.

Now, our aim is to do a GWAs meta-analysis with the UK Biobank data, including all patients with stroke attributed to intracranial stenosis or occlusion in order to increase the sample size and to identify new polymorphisms associated with this disease. Furthermore, the polymophisms identified will be further studied to find biological implications.

During reproduction, a parent passes on a single copy of a given gene: either a copy from their mother, or a copy from their father. However, what are referred to as “selfish” genetic elements can stack the deck in their favor, increasing odds of being passed on. While selfish genetic elements are described for a variety of organisms, their evidence in humans has been inconclusive. We aim to test for the presence of such non-randomly inherited genetic elements in the human genome, particularly with respect to variation predisposing a parent to having boys versus girls. Researching such “selfish” elements can inform us of the role genes play with regard to difficulty conceiving, sterility, miscarriage, and developmental disorders.

There are multiple examples in animal systems demonstrating how the inheritance of sex can be biased from the expected 1:1 ratio. One mechanism leading to sex ratio bias can be unequal transmission of sex-determining chromosomes. For example, a man who produces more mature sperm bearing Y chromosomes than X chromosomes would present as a bias toward having sons. Alternatively, sex ratios could be biased due to differences in the survival (or viability) or male or female embryos. A mother passing on an X chromosome mutation that makes it less likely for male embryos to implant or survive through pregnancy, for example, would manifest as a bias toward having daughters. One specific case of reduced viability, the “Mother’s Curse” hypothesis, predicts an accumulation of mitochondrial mutations that only are harmful to males.

We will test for the above biased transmission mechanisms, leveraging the sample size within the UKBiobank. By investigating genetic factors related to biased sex ratios, this project aims to shed light on heritable factors related to sex-specific developmental disorders or sterility. This project should be completed within two years from the date of data acquisition.

It is widely recognised that early childhood trauma (including neglect) is strongly associated with later difficulties during adolescence and adulthood, including mental health problems and poor social and health outcomes. Some people, however, seem to be resilient to early adversity, and despite trauma or neglect during their childhood manage to have a good quality of life in adulthood. We know that psychological characteristics associated with such resilience include positive emotions, optimism and adaptive coping strategies. Other factors, such as good parenting and social support, particularly close, confiding relationships, can engender resilience, particularly through a ‘buffering’ effect on life events and adversity.

Other research has also focussed on biological factors that might influence how resilient a person is. These include hormonal factors as well as the body’s ‘stress response’. Genes are also believed to be important, and by studying the coding regions of key hormonal and stress response genes, some small studies have shown that differences in the coding regions of these genes are associated with how resilient a person is.

As there are probably many genes that are important, looking at the association between variation in a person’s DNA and their resilience should examine all genes and the DNA between genes. This is undertaken using a method called genome-wide association (or GWA), which is the preferred option for studying the genetics of any complex trait. We now plan to study resilience in this way using the UK Biobank, a population-based cohort of ~500,000 participants recruited in the UK. Phenotype information pertaining to early adversity and later adaptation have been collected on participants in addition to their DNA. The aim of this study is to undertake a GWA study of resilience to identify markers (polymorphisms) that are associated with a person’s resilience. In this way, we hope to identify some of the polymorphisms that are associated with resilience as well as the genes that are important and their pathways and functions in the body.

Tinnitus is a distressing and persistent condition in a wide portion of the population, but little is known about its functioning and direct cause. Without a solid foundation of understanding, it becomes incredibly difficult to develop effective treatments. In order to develop drug therapies for tinnitus, then, it is critical to understand its mechanism of action. To do this, we plan to conduct a genetic study based on data from both the UK Biobank and the U.S. Department of Veteran Affairs to identify genes impacting the development and severity of tinnitus. Then, by relating these genes back to their corresponding pathways and validating the change in activity of these pathways in between healthy and tinnitus-affected individuals, we can identify potential targets for tinnitus therapies. Using a data-based drug screen, it is possible to identify compounds that change the activity of these targets, and by extension, alter the effects of tinnitus in an individual. We are then able to test these compounds in a mouse model that corresponds to the cause of tinnitus, which may be better representative of the effectiveness of the drug against a known cause of tinnitus.

The project aims at the development of novel medicines for diseases that are related to structures, so-called receptors, which are located on the outside of cells in the human body and that can bind certain substances and thereby convey signals to the inside of the cell. In this project we focus on a particular kind of receptors, the “G-protein coupled receptors” (GPCRs). GPCR receptors are important for conveying signals originating from nerves, hormones, and the immune system, which are all important for keeping the body in a balanced healthy state. The central role of receptors in signalling makes them important drug targets. However, only a minority of GPCR receptors are currently used in medical treatment, because our knowledge about them is incomplete.

In this project, which we estimate to take about three years, we would like to use differences in the genetic information among people, and to link their medical records to identify receptors that could be involved in diseases.

About 60 new changes per generation are added to a person’s genetic material (DNA), and over time many of those changes — called mutations — have also affected GPCR receptors, as these receptors are encoded by the human DNA sequence.

The GPCR mutations that we will see in healthy people in this project will enable us to make modified GPCRs with normal function. The GPCR mutations that turn out to be related to diseases will be engineered to study their modified function.

Most diseases can be traced back to a genetic origin, often involving multiple mutations. By analysing the genetic information of many people, we can determine which mutations are associated with groups of people who are affected by a disease compared with groups of healthy people. GPCR receptors carrying disease-related mutations can be made and tested for their biological function or interaction with potential drug molecules. We intend to study diseases related to inflammation (for instance, gut inflammation) and neurodegeneration, which is the irreversible loss of nerve cells, leading to dementia. Alzheimer’s disease is the most common cause of dementia, and therefore it is important to find treatments to cure it. The relatively large number of people present in the UK Biobank (or their relatives) who are affected by Alzheimer’s disease also means that we can fruitfully use the data collected in the UK Biobank to relate GPCR receptors to neurodegeneration.

Autoimmune diseases, such as multiple sclerosis, type 1 diabetes, coeliac disease, rheumatoid arthritis, lupus and thyroid disease, are common health problems in the UK. Comparing and contrasting the results from disease specific GenomeWide Association Studies (GWAS) has shown that there are common aetiological mechanisms underlying these diseases and in this context we are proposing to undertake a GWAS with autoimmunity as the primary endpoint using genotype data from the UK Biobank as controls. Each disease specific group in this collaborative application has genotype data from many thousands of affected UK individuals. Imputation will enable us to compare these existing data with UK Biobank subjects. To supplement these already extensive existing data we will be applying for funding to undertake genotyping of the UK Biobank chip in several thousand previously untested cases. This GWAS would be well powered to identify even low frequency risk variants. The results generated might provide invaluable insight into aetiology and thereby enhance the potential development of safe, effective, rational therapies.
The main analysis would exclude those participants with self-reported autoimmune disease at baseline. Further analyses would be done using self-reported cases and controls from UK Biobank in combination with data from other studies. Disease specific analysis will also be performed with similar treatment of the relevant self reported cases. It is anticipated that most if not all of the untyped UK autoimmune disease patients are not currently part of UK Biobank. Full Cohort

Our motivation for studying the UK Biobank data set is based on the observation that traditional detection methods for finding indicators of a particular disease state are not well translated with ethnic minorities. The intention is to develop algorithms that can learn genetic characteristics from large datasets, and then develop transfer learning methods for smaller datasets, ie. Understudied populations. We intend to deal with population specificity, with a particular focus on the population of the UAE.
For benchmark purposes, we aim to study diseases, namely coronary artery disease (CAD), atrial fibrillation, type 2 diabetes, inflammatory bowel disease, and breast cancer. The UK Biobank comprises a dataset from ethnically diverse participants, which is a crucial requirement for our purposes.
In particular, we will first perform population stratification on the dataset, clustering ethnic minorities. Our algorithms will then be trained on a very large dataset (representing the majority) while leaving out one or more ethnic minorities. We then apply Artificial Intelligence to the left-out minority. In a final step, we are interested in applying the established algorithm to local datasets.

Research question
What is the frequency and phenotypic spectrum associated with deleterious variants in genes that are known to cause monogenic disorders in UK Biobank, and what are the genetic and environmental factors that influence disease expression?

We aim to address this question for the genetic disorders that are the focus of translational and fundamental research and patient care in the Leiden University Medical Center (LUMC) in the Netherlands. These LUMC-expertise monogenic disorders include: hereditary stroke and dementia syndromes, chromatinopathies, hereditary cancer, muscle disorders and DNA-repair disorders.

Objectives
To utilize linked genetic and phenotypic information in UK Biobank in order to:
1) determine the frequency of deleterious variants in genes associated with LUMC-expertise monogenic disorders
2) determine the phenotypic spectrum associated with these deleterious variants
3) identify genetic and environmental modifiers that play a role in disease expression

Data from UK Biobank will be combined with data from LUMC in-house patient registries, and with disease-relevant external databases. In this way, the full phenotypic spectrum is included, thereby increasing the power for the detection of modifying factors.

Scientific rationale
With the increasing use of whole exome and whole genome sequencing in routine clinical care, milder phenotypes associated with variants in genes known to cause severe monogenic disorders are increasingly recognised and identified. This creates challenges for patient and family counselling, but also opportunities for understanding disease pathomechanisms and identifying modifiers. An example is the research on NOTCH3 variants performed by the LUMC under UK Biobank application no. 74162, which has resulted in improved disease prediction in the clinic through the identification of a strong genotype-phenotype correlation (Rutten et al. Neurology 2020; Hack et al. Brain 2023).

Even in the absence of neurodegenerative disorders like dementia or Alzheimer’s disorder, aging is associated with changes in the brain as well as deficits in cognitive ability. However, brain aging does not occur at the same pace or in the same way for all individuals. Some may age “optimally” with little to no changes in memory or cognition, while others may experience severe forms of impairment. This project aims to identify factors that are most important for maintaining brain health through late adulthood. We will study individual differences in genetics, lifestyle, and environment in combination with measures of brain shape, size and function. By characterizing which factors are most important for promoting longevity, this work will help craft public health policy and identify specific biological targets for medical research and intervention.

The goals of this project are to develop a list of DNA mutations linked to diabetes and obesity, which will be used to develop a personalised nutrition platform. Indeed, with the list of mutations, we can assess the potential risk a certain person has to develop these particular diseases. This risk scoring will be combined with clinical data as part of clinical studies to assess the effect of diet personalisation on young people with diabetes and obesity. The UK biobank data will be used to refine the list of DNA mutations and to develop the risk score models prior to the clinical studies.

The expected impact from this project is to motivate healthy dietary choices. This will be done by providing personalised nutritional advice, based on each person own information (DNA, biomarkers, clinical data!). Recent studies are showing that two persons can have a very different response to the same diet. This could explain why public health campaigns have not been effective in halting the rise in nutrition-related non-communicable diseases (NCDs) around the world. Notably because they rely on the “one size fits all” approach, which fails to take into account how each person will respond to a given diet. We expect personalised nutrition to provide a solution to this issue.

Aim: We aim to add real-world evidence to our existing database, expanding relationships between different biological entities and, where applicable, assign a value to these relationships, which would inform us as to whether or not the relationship predicts a disease.

Scientific Rationale: Biological databases, generally, are created in isolation of one another by individual research groups. Consequently, Biological databases do not include relevant interactions with other biological databases. We extract information from separate databases, standardize it, creating one interconnected database. By linking related entries from these databases together, we have identified relationships that would otherwise not be captured if the databases were used independently, creating a web of related biological concepts. UK Biobank is an important data source of numerical experimental information on phenotypes, biomarkers, and diseases, which would improve the quality of the information provided by the relationships in our database.

Public Health Impact: The addition of real-world evidence to our existing database would provide important information on protein expression and its relationship to normal vs. disease states not available in traditional biological databases. This information will help pharma companies better understand the biology and molecular mechanisms for their target of interest. Further, our internal report showed that as much as 80% of preclinical experiments failed because of poor experiment design; the new network that we aim to develop will play a direct role in helping scientists execute better experiments.

Project Duration: The project duration will be 36 months.

Stroke is a leading cause of death and disability in the UK. Much of stroke management revolves around addressing risk factors with medications and lifestyle modification. However, 30% of those who suffer stroke have no known risk factors. Hence, there is need to better identify individuals who are at high risk of stroke, and particularly those where the benefit of treatment outweighs the risk.
ABSTRACT is a three phase study that looks to address this issue by (1) using artificial intelligence (AI) to predict stroke risk from routine hospital data, (2) to validate this model on external datasets, and (3) validate the ability to improve outcome by guiding clinical decision making.
Phase 1 has shown promising results, with Xgboost achieving an AUC of 94% when using routine blood test data, medical history and CT/MRI head imaging data to predict future stroke risk from a cohort of 9155 stroke cases and 109,875 controls in Southwest England. The model also identified several novel risk factors for stroke, such as liver function tests and C-reactive protein.

We now look to commence phase 2 of our project and validate these findings on an external dataset. At this time we ask the following research questions:
1. Can AI accurately predict stroke risk from routine hospital data?
2. How well does our model generalise to UK Biobank participants?
3. Does model performance vary across demographic and clinical subgroups?

Based on these questions, our research aims and objectives are therefore as follows:
1. To perform external validation of our stroke risk model using UKbiobank data.
2. To validate the novel risk factors for stroke identified by our models in phase 1.
3. To Assess model performance and calibration across different demographic subgroups

Following the results of this external validation we will then look to commence phase 3 of our project and assess the effectiveness of models in guiding clinical decision making and stroke risk management.

1. Integration of Multi-Omics Data and Exploration of Kidney Disease Mechanisms:
A comprehensive multi-omics analysis leveraging brain MRI, carotid ultrasound, and proteomic, genomic, and metabolomic profiles to identify risk factors and predictive markers for kidney disease and cardio-cerebrovascular events, with an emphasis on the pathophysiology of cardiovascular-kidney-metabolic (CKM) syndrome.
2. Investigating Genetic and Environmental Risk Factors for Kidney Diseases:
Combine UK Biobank genomic data (whole-genome sequencing, genotyping) with environmental exposure data (air pollution, lifestyle factors, etc.) to analyze how gene-environment interactions impact the development of kidney diseases, such as chronic kidney disease and acute kidney injury.
3. Develop an Early Warning Model for Kidney Disease:
Utilize longitudinal data from UK Biobank (including biochemical markers and imaging data) along with electronic health records (linked via the NHS) to build risk prediction models based on machine learning or deep learning techniques.
4. Evaluate the Long-Term Effects of Medications and Treatment Strategies:
Analyze medication records and follow-up data from UK Biobank to assess the impact of renin-angiotensin system inhibitors (ACEI/ARB) and novel targeted therapies (e.g., rituximab) on the progression of kidney diseases.

Through comprehensive multi-dimensional data analysis, we will deeply explore the pathogenesis of hemorrhagic stroke and develop an early warning model:
1. Integrate genomics (whole genome sequencing, exome), proteomics (plasma protein measurement), metabolomics (serum metabolites) and neuroimaging data (brain MRI/CT) to reveal the molecular network of hemorrhagic stroke. This will help identify new biomarkers and clarify the interactions between genetic susceptibility, vascular inflammation pathways, coagulation dysfunction, and blood-brain barrier integrity.
2. Combining genomic data (whole genome sequencing, genotyping) and vascular risk factors (hypertension, lifestyle, dietary patterns) to construct a polygenic risk score (PRS). PRS can quantify the combined effects of genetic variants (such as APOE, COL4A1 gene mutations) and environmental factors on hemorrhagic stroke risk and help identify high-risk populations for preventive interventions.
3. Using brain imaging data (MRI quantification of cerebral small vessel disease, microbleeds detection) and longitudinal health records (blood pressure monitoring, coagulation parameters), develop a deep learning-based system to automatically identify imaging biomarkers of hemorrhagic stroke risk. Combined with clinical indicators, a dynamic risk prediction model is created to detect potential hemorrhagic events early for timely intervention and improved patient outcomes.
In general, I will comprehensively analyze the pathogenesis of hemorrhagic stroke, identify new biomarkers, and develop early warning models through multi-omics mechanism analysis, gene-environment interaction research, and neuroimaging phenotype analysis to provide a scientific basis for clinical prevention and treatment.

Through multi-dimensional data analysis, we will deeply explore the pathogenesis of pulmonary fibrosis and develop an early warning model:
1. Integrate genomics (whole genome sequencing, exome), proteomics (plasma protein measurement), metabolomics (serum metabolites) and imaging data (chest CT/MRI) to reveal the molecular network of pulmonary fibrosis. This will help identify new biomarkers and clarify the interactions between genetic susceptibility, inflammatory pathways and metabolic abnormalities.
2. Combining genomic data (whole genome sequencing, genotyping) and environmental exposure data (air pollution, lifestyle, etc.) to construct a polygenic risk score (PRS). PRS can quantify the combined effects of genetic and environmental factors (such as MUC5B gene mutation and smoking) on !!the risk of pulmonary fibrosis and help identify high-risk groups.
3. Using chest imaging data (CT quantification of interstitial lung changes) and longitudinal health records (pulmonary function, hospitalization data), develop a deep learning-based system to automatically identify imaging biomarkers of pulmonary fibrosis. Combined with clinical indicators, a dynamic risk prediction model is created to detect pulmonary fibrosis early so that timely intervention can improve patient prognosis.
In general, I will comprehensively analyze the pathogenesis of pulmonary fibrosis, identify new biomarkers, and develop early warning models through multi-omics mechanism analysis, gene-environment interaction research, and imaging phenotype analysis to provide a scientific basis for clinical diagnosis and treatment.

Research questions: Metabolic-associated steatosis liver disease (MASLD), as the new name for non – alcoholic fatty liver disease (NAFLD), still requires a comprehensive elucidation of its complex pathophysiological mechanisms, diverse clinical phenotypes, and the synergistic effects with metabolic diseases such as cardiovascular diseases and type 2 diabetes. In large-scale population cohorts, how to accurately identify susceptible populations, make early diagnoses, and effectively intervene to block disease progression remains a key scientific issue that urgently needs to be addressed. How does the genetic basis of MASLD affect its occurrence and development? What is the interaction mechanism between different metabolic disorders (such as obesity, diabetes) and MASLD? Do significant differences exist in the clinical characteristics and prognosis of MASLD patients under different BMI classifications (e.g., lean vs. overweight/obese)? What are the effectiveness and limitations of existing diagnostic markers and
treatment strategies in the real world?
Research objectives: This study aims to utilize the large-scale, multi-dimensional, and prospective population cohort data of the UK Biobank to deeply analyze the pathological mechanisms, clinical characteristics of Metabolic-associated steatosis liver disease (MASLD), and its association with cardiometabolic diseases. On this basis, potential diagnostic and treatment strategies will be explored.
Scientific rational: MAFLD involves complex pathological mechanisms centered on hepatic steatosis, linked to insulin resistance, inflammation/oxidative stress, lipid metabolism disorders, and gut-liver axis dysfunction. Clinically, it ranges from asymptomatic to advanced liver disease, associates with metabolic diseases, includes lean cases (BMI<25), and increases cardiovascular risk and sarcopenia likelihood.

We will use multiple UK Biobank resources to better understand the relationship between both genetic and environmental factors with cardiovascular diseases including congenital heart disease. Over the course of the project we aim to discover novel associations influencing disease risk, the reasons why some other diseases occur together with cardiovascular disease, and discover factors that influence the severity of disease over time. This could ultimately impact future healthcare as it may enable earlier diagnosis and treatment in families affected by these conditions, improving the development of personalised care pathways.

Scientific rational for research
Evidence links several pregnancy-related conditions to long-term health risks, including gestational hypertension and cardiovascular disease, gestational diabetes and metabolic disorders, and perinatal mental health issues with ongoing psychological needs. Despite these well-established associations, there is limited understanding of how pregnancy complications translate into long-term health trajectories and when intervention is most effective.
More precise models for risk prediction, incorporating a range of health and ‘omic factors, could enable earlier identification of high-risk individuals. By leveraging pregnancy as a predictor of long-term health, there is an opportunity to improve preventative strategies, personalise care, and enhance health outcomes across the life course.

Research question
Can the utilisation of ‘omics data incorporated alongside traditional clinical risk factors predict future ill health after pregnancy complications?

Objectives
1. Assess long-term disease risk following pregnancy complications
– Evaluate the progression from gestational diabetes to type 2 diabetes
– Investigate how hypertensive disorders of pregnancy (e.g., preeclampsia, gestational hypertension) influence future cardiovascular disease risk
2. Examine the impact of pre-existing conditions on pregnancy-related disease progression
– Determine how pre-existing cardiometabolic conditions (e.g., obesity, chronic hypertension, type 1 or 2 diabetes) during pregnancy influence long-term health trajectories
3. Utilise longitudinal clinical and biochemical data to define disease trajectories
– Establish patterns of onset and progression of diabetes, cardiovascular disease, and related metabolic conditions
4. Leverage multi-omics data to explore genetic and molecular determinants of disease
– Use whole genome sequencing, exome sequencing, and proteomic data to investigate genetic drivers of metabolic and cardiovascular disease

Cardiomyopathy is a disease of the heart muscle. It causes the heart to have difficulties to pump blood to the whole body, which can lead to symptoms of heart failure. Cardiomyopathy also can lead to sudden cardiac death. In China, it is estimated in 2023 that there are about 1,000,000 patients suffering hypertrophic cardiomyopathy and about 120,000 patients suffering dilated cardiomyopathy. According to “2024 heart disease and stroke statistics: a report of US and global data from the American Heart Association”, 410,000 deaths were estimated for cardiomyopathy and myocarditis based on 204 countries and territories in 2021.
Since genetic testing is costly and has a long report generation period, the phenotypic cardiac MRI (CMR) imaging is usually considered as the necessary examination utility for cardiomyopathy diagnosis, as it can be used to measure the cardiac structure and function, evaluate the severity of cardiomyopathy. However, the analysis of cardiac MRI imaging is a heavy workload to physicians as it requires a lot of time to contour heart anatomy and requires long time training to improve their accuracy. For the similar reasons, only two commercial software tools are available over the world for CMR analysis for a long time.
Beyond the CMR imaging, clinical data such as baseline and lab test variables are also important for cardiomyopathy diagnosis. So given the amazing progress in deep learning techniques, we aim to apply deep learning on multi-modal medical data to diagnose various forms of cardiomyopathies.
However, in the medical application scenario, large deep learning models are inherently black boxes, and hard to gain the trust of medical professionals. We break the large black-box into 3 pieces and have medical knowledge embedded in the framework, so that we not only take advantage of the deep learning advances, but also provide an opportunity for medical professionals to trust AI.
Furthermore, our knowledge oriented deep learning framework achieved high diagnosis accuracy to benefit cardiomyopathy patients and could run in portable or small computing devices.
Through our research project, the public attention may be directed to other possibilities than the overwhelming large AI models, especially in the medical AI region. Our knowledge oriented approach perfectly avoid the illusion pitfalls that most large AI models currently face.

This research dives into the development and progression of chronic diseases, including, but not limited to, cardiovascular disease, respiratory disease, metabolic disease, and cancer. It aims to 1) track how individuals transition from healthy states to single chronic disease, then multimorbidity, and ultimately death. 2) identify distinct patterns in this progression, considering which diseases co-occur and the order of their appearance. 3) investigate the causal relationships between lifestyle factors, genetic predispositions, and the development and progression of chronic diseases. Identify potential mediators and modifiers that influence these pathways. 4) assess the prevalence and impact of overdiagnosis in chronic diseases, including the identification of factors contributing to overdiagnosis.

CHIP is an age-related phenomenon characterized by somatic mutations in hematopoietic stem cells. These mutations lead to the clonal expansion of mutant blood cells without overt hematologic malignancies. While CHIP is primarily associated with an increased risk of atherosclerotic cardiovascular diseases (ASCVD), emerging evidence suggests a potential role in metabolic disorders, including type 2 diabetes mellitus (T2DM) and related complications. Prior studies on CHIP in relation to T2DM have demonstrated an increased prevalence and incidence of T2DM. Although there have been individual cohort studies on the risk of CHIP on developing diabetic complications, evidence from large cohort studies is lacking. The increased risk and mortality pertaining to ASCVD, we hypothesized microvascular complications, especially diabetic kidney disease and its associated phenotype of albuminuria and proteinuria, to be influenced by the aberrant immunologic response in CHIP carriers with diabetes.
We thus aim to analyze individuals with whole-exome sequencing data to compare the incidence of diabetes-related complications (e.g., nephropathy, retinopathy, neuropathy) between CHIP carrier T2DM individuals with non-CHIP carriers T2DM individuals, and baseline non-diabetics with or without CHIP who developed T2DM during follow-up

1.Research question
The diagnosis of psoriasis faces multiple challenges. Its skin lesions are highly similar to seborrheic dermatitis and eczema, easily causing confusion and misdiagnosis. Early symptoms are mild and subtle, often overlooked and delayed. Lacking specific biological indicators, diagnosis relies on clinical manifestations and historical experience. With significant symptom differences among subtypes and dynamic changes in lesions during progression, difficulties increase, demanding specific biomarkers to enhance diagnostic efficacy.
2.Objectives of the study
To construct a 10-year psoriasis risk prediction model based on metabolomics for psoriasis risk stratification.2.To explore the predictive value of key predictors for the risk of psoriasis.
3.Scientific basis
Plasma metabolites can directly reflect the physiological and pathological status of the organism, with dynamic sensitivity and easy detectability, providing specific molecular markers for diseases; meanwhile, machine learning excels at mining potential associations from massive high-dimensional metabolomics data, overcoming the limitations of manual analysis to improve the efficiency of marker screening and diagnostic model construction.

The diabetic and hypertensive-related nephropathy is a serious life-threatening condition. Catching the condition early and taking actions in advance can help slow down or even stop the progression of these diseases. Diabetic DN and hypertension are affected by many genetic changes, and non-genetic features frequently mentioned as social and environmental factors. A polygenic risk score is a score that can estimate a person’s risk for developing a disease based on their genetics or DNA. This is often displayed in a way that tells how at risk a person is for a specific condition. The proposed study is to investigate these changes and factors to understand the role that genetic and environmental factors play in developing diabetic DN and hypertension across different populations. The expected value of this study is to produce a novel scoring system to predict the risk of developing diabetic DN and hypertension in diabetes patients, based on the total number of changes and other information that is related to this condition. Our goal is to provide an accurate risk assessment tool that can help diabetes patients to manage their personal risk for developing diabetic DN and hypertension. The proposed duration of this study is 2 or 3 years.

Latest advances in machine learning, specifically the deep learning technique have shown tremendous potential in medical applications for diagnosing and detecting diseases at early stage based on imaging data. In this research project, we aim to make use of these advances to automatically identify people at risk of certain eye and systemic diseases such as Diabetic Retinopathy, Age-related Macular Degeneration, Stroke and Cardiovascular diseases at early stage and help prevention. For example, the number of Americans with diabetic retinopathy is expected to rise from 7.7 million to 14.6 million between 2010 and 2050. It is estimated that nearly 80% of these cases can be preventable. Similarly, By 2050, the estimated number of people with AMD is expected to more than double from 2.07 million to 5.44 million. Early detection is key to prevent these diseases and reducing the unnecessary overhead of medical bills, and save sight and lives. The project will make use of the vast imaging and demographic data that UK Biobank has, to build such systems to predict the onset of these diseases. We estimate that it will take up to two years to complete the research project.

Our project uses modern machine learning techniques to predict whether people will develop certain diseases, specifically focussing on chronic pain, such as diabetic polyneuropathy, Chronic Regional Pain Syndrome (CRPS) and chronic lower back pain. We aim to use this platform to help understand the reasons why people develop those conditions and to inform further research into candidate mechanisms through which chronic pain occurs. By using the large sample size available through the UK Biobank (UKB, n > 500,000), we will determine for example whether certain blood biomarkers are co-incident with chronic pain onset, which genes are most highly associated with chronic pain, and what is the intersection between genetics, lifestyle, biomarkers and brain structure in predicting chronic pain conditions.

This research supports a movement towards more personalised medicine by helping researchers develop a deeper and more bespoke understanding of which individuals are at higher risk of chronic pain based on clustering populations according to both genetics and phenotype. Through this research we hope to cast light upon how different aspects of pain affect different people, based on their genetics, biometrics, brain scans and many other features.

The UKB is unique in the sense that it contains multi-modal data for a range of variables across a large number of individuals. This means that for the first time it is feasible to build models that look across thousands of variables without pre-supposing any particular association, to uncover novel risk factors for a disease of interest. As a result of this it will be possible to firstly validate the existing factors responsible for a disease, but in addition to both identify new factors and uncover more complex connections between different factors, for example the importance of certain inflammatory biomarkers in predicting chronic pain.

Raising HDL, using medication, has not been shown to improve cardiovascular outcomes, despite epidemiological studies showing that high HDL is beneficial. A recent genome-wide-association study has shown that the HDL raising drug, dalcetrapib, reduces cardiovascular events and atheroma in patients with a particular genotype. This effect may be either gene-drug or gene-HDL related. We propose a mendelian randomisation study to evaluate the relationship between de novo HDL levels and the ADCY9 rs1967309 and its influence on carotid artery intima media thickness (CIMT), as a marker of atherosclerosis. UK Biobank’s stated purpose is to support health-related research that is in the public’s interest. Our study will have wide relevance to population health given the high prevalence of atherosclerosis, and morbidity/mortality associated with the disease. If a relationship is shown to exist between HDL and the ADCY9 gene, which is independent of dalcetrapib, then this drug-gene association will be more widely relevant to the population, even in the absence of this drug. Both the drug and the gene test are patented. No biological samples in the UK Biobank will be used in this study. Subjects will be identified by their HDL levels. Subjects who have undergone CIMT testing will be stratified by their HDL levels. Subjects in the lowest and highest HDL quartile will be evaluated to test the relationship between CIMT and the ADCY9 gene. This project can be completed with only electronic data and will not require consumption of biological samples. The CIMT subset of the full cohort will be used. This study will only be possible if either the ADCY9 rs1967309 SNP has been tested on the UK Biobank array or can be imputed from the existing data.

Aims:
We aim to investigate and clarify the direct links between health habits and their effects on cardiovascular and renal risks. Using the Mendelian randomization method, we hope to gain a clearer understanding of these connections, which traditional studies might miss due to possible biases.

Scientific Rationale:
It’s often challenging in contemporary health research to clearly determine the direct links between habits and health outcomes. Our approach is different because we use Mendelian randomization, a method that uses genetic differences as tools, to identify direct causes without the usual interfering factors. We plan to use the detailed data from the UK Biobank, especially the GWAS data, to find specific genetic markers linked to health habits.

With this genetic information and a wide range of data on diet, health, and background factors, we can create studies that mimic randomized controlled trials. This will help us estimate the direct effects of health habits on heart and kidney health outcomes. The broad and detailed nature of the UK Biobank data means our analysis can consider a wide range of diets, backgrounds, and health outcomes.

Project Duration:
We expect the project to last three years. The first year will be about gathering data and an initial analysis. The next year will focus on the Mendelian randomization and more detailed studies. The last year will be for interpreting the data, confirming our results, and sharing our findings.

Public Health Impact:
The results of our study could be very important for public health. With growing concerns about heart and kidney diseases, it’s essential to understand the direct links between our daily habits and these health problems. Using the detailed UK Biobank data, we aim to give more specific and evidence-based health advice.

By creating health improvement programs and personalized recommendations based on our findings, we hope to improve public health and well-being. This study could also guide future health research by showing the importance of using wide-ranging and detailed data for creating evidence-based health advice.

In Conclusion:
For our study to be successful, we need to access various types of information from the UK Biobank. This request highlights the importance of having a wide range of data. It shows that a thorough approach, based on detailed and varied data, can produce strong health research that can be applied in many ways.

We propose to run a large scale analysis of genome-wide association studies (GWAS) to identify novel common, low-frequency and rare genetic variants associated with vitamin D (specifically 25-hydroxyvitamin D (25OHD)). We are aiming to expand the understanding of how genetic variation contributes to levels of vitamin D which is known to be associated with a host of different disorders and also with biological pathways important for diseases such as cancer. This understanding will not only help us to understand what determines levels, but also apply this knowledge to assess the impact of vitamin D levels on disease. Vitamin D insufficiency affects up to half of otherwise healthy adults with detrimental impacts on public health. Approximately 50% of the variability in 25OHD levels is attributable to genetic factors. Although genetic factors are speculated to contribute substantially to this variability, the identified to date four common variants explain little of the heritability. The objective of the present study is to detect additional common, low-frequency and rare variants in novel genetic loci of large effect, associated with 25OHD levels, enabling the identification of novel pathways implicated in 25OHD metabolism and groups of individuals at risk for 25OHD insufficiency. We have collected 12 participating studies, with approximately 30,000 individuals with genome-wide genotypes and 25OHD levels. All cohorts have assessed the effect of genetic variants on 25OHD levels. We have imputed each cohort to the UK10K/1000Genomes reference panel which enables more accurate estimation of low frequency and rare genotypes. While we have identified preliminarily interesting findings, we aim to include the UKBiobank data to replicate our findings and identify potential novel loci. To do so, we will undertake a fixed-effects meta-analysis of all cohorts of the effect of each genetic variant on 25OHD levels. For these analyses we request access to genome-wide genotyping data, serum levels of 25OHD, and the following covariates: sex, age, body mass index, date of vitamin D measurement and reported vitamin D intake through food and supplemental sources. Specifically we request the above data for the entire UKBiobank cohort as and when available. We would like to use and extend (with the full release when available) imputed GWAS data generated as part of UKBiobank project 8786. This will greatly speed up our project and prevent replication of effort and minimise the amount of storage we use for this project.

The proposed study aims to address three research questions:

* Are there genetic defect patterns common to broad categories of disease such as all cancers, all psychiatric disorders, or all musculoskeletal disorders?
* Conversely, are there genetic patterns that help protect people against broad categories of disease, such as cancer or cardiovascular disease?
* Are there genotypic variant signatures allowing stratification of patients that could inform the risk of developing a disorder and likelihood of drug therapy response?

The proposed research would improve our understanding of the genomic basis of disease formation (or disease prevention). We hope to identify individual genetic defects or combinatorial defect clusters that are commonly associated with broad categories of disease, that is, found in significantly higher numbers of patients compared to healthy controls. Similarly, we hope to identify protective signatures that are found in many more healthy controls compared to afflicted individuals.

In addition, we hope to develop new improved ways of identifying patients at risk of developing a disease or its complications, and enable patients to be treated with a drug therapy regimen that is tailored to their individual needs. Successful results would help future researchers identify means to increase human longevity and wellness by manipulating genetic mechanisms involved in broad categories of disease. Through follow-on studies, researchers may identify new drugs that work across broad categories of disease. Such drugs with broad applicability may cost less to develop, test, and bring to market, thus helping everyone afflicted with those diseases. Tailoring therapies to patients’ individual needs may significantly reduce the burden on the healthcare system through reduced side effects due to drug interaction or lack of therapy response, leading to hospital admissions.

The project duration is 24 months.

Aims- Aim 1. To develop a metabolite profile of muscle function in a cross-sectional analysis of middle and older aged adults. Aim 2: To validate metabolite profiles of muscle function that are predictive of muscle function change over time.

Scientific rationale- Sarcopenia is a condition that may occur with aging that results in muscle weakness and loss. Weak and smaller muscles are linked to an increased risk in disability, chronic disease and mortality. Sarcopenia is treated with physical activity and exercise. The focus only on activity suggests that each person can control whether their muscles get smaller and weaker. We believe that muscle strength and size are not only impacted my activity, but metabolism as well. Our goal is to identify metabolic markers from people who appear healthy but demonstrate either poor, average, or good muscle function. If metabolic disturbances can be identified during middle age, it can then help prevent changes with aging. Muscle function is difficult to fully characterize given 600 muscles uniquely responding to lifestyle choices, aging, disease, pharmacology and inter-organ and tissue crosstalk such as with liver, bone, and heart. Therefore, we are utilizing a systemic approach by analyzing blood metabolites of individuals with above average, average, and below average muscle function between age categories of 40-59 and >60 years old. The metabolic profile will identify those who are at risk and should be intervened at an earlier timepoint in future studies. Collectively, this data will establish multiple proposals that will identify those who are at long-term risk for muscle dysfunction and to personalize exercise prescription in future studies.

Expected duration of project- 2 years

Public health impact of the work- Identifying metabolic contributions to poor muscle health can provide direction on how to best manage physical activity and other factors such as nutrition and nutraceuticals.

I. Research Questions
1. How can we develop a metabolomics-specific foundation model to capture the complex relationships between metabolites and aging/disease outcomes?
2. Can a foundation model effectively predict biological age and disease risk by integrating multimodal data, including metabolic profiles and environmental factors?
3. What are the key metabolic pathways driving population-specific aging patterns, particularly influenced by diet and environmental exposures?

II. Objectives
1. Develop a metabolomics-specific foundation model to harmonize heterogeneous data and capture complex aging patterns.
2. Fine-tune the model for biological age estimation and disease risk stratification using UK Biobank data.
3. Identify key metabolic pathways driving population-specific aging patterns using interpretability techniques.

III. Scientific Rationale
1. Data Heterogeneity: Existing metabolomics studies are hindered by data heterogeneity. A foundation model can harmonize diverse datasets, improving generalizability.
2. UK Biobank Data: The UK Biobank’s extensive longitudinal data provides a unique opportunity to develop a robust model for aging and disease prediction.
3. Clinical Relevance: Fine-tuning the model for biological age and disease risk will provide actionable insights for early detection and personalized interventions.

IV. Plan for Disseminating Our Findings
We will disseminate our findings through high-impact SCI publications and presentations at international conferences to engage the academic community. In addition, we will develop and release an open-source web tool based on our findings, enabling the public and researchers to explore and apply the model. Our goal is to make our results accessible through both academic and public channels, ensuring broad impact and contributing to the field.

Aging is commonly associated with an increase in the risk of complex phenotypes onset, such as cardiovascular diseases, diabetes, cancer, etc. One can consider healthy aging or mortality as complex phenotypes shaped by susceptibilities to many underlying diseases and with significant contribution of environmental factors. In this context, the outcomes could be defined as age of major disease onset, age of critical health deterioration or age of death. We will use such interpretations to build a predictor for the individual risks of various aging outcomes. Specifically, we will start with most common diseases of aging – cardiovascular disease (and incidents) and cancer. Further, the mathematical design of the predictor will be investigated in other phenotypes – diabetes, autoimmune disorders, to understand specific features required for each disease type.

Main deliverables are: first, information about feature importance, such as, selection of data points that are most informative for health outcome. Second, average health endpoints for individuals with similar input parameters, found in training data. And finally, data-based individual longevity recommendations for managing parameters with the highest contribution to age/mortality prediction. This work is in line with the UK Biobank’s aim of enabling research to improve prevention, diagnosis and treatment of illness and the promotion of health throughout society.

Our aims are to investigate physical activity behaviour and sedentary behaviour, and cancer risk, including better estimations of the magnitude of risk by different cancer sites. Specifically, we aim:
1) To evaluate correlations between physical activity and sedentary behaviour data and relevant biomarkers related to physical activity/sedentary behaviour and cancer
2) To analyse the association between physical activity and sedentary behaviour data and relevant genetic variants identified within GWAS data and cancer risk.
3) To analyse factors related to the built environment, physical activity, sedentary behaviour, and cancer risk.
The aim of UK Biobank is to improve the prevention, diagnosis and treatment of serious and life-threatening illnesses, including cancer. The proposed research will elucidate the role of physical activity and sedentary behaviour in cancer prevention. The study will provide insight into biological mechanisms by which physical activity and sedentary behaviour may impact on different cancer sites, the role of individual modifiable and non-modifiable characteristics, and built environment factors in influencing the association between physical activity, sedentary behaviour, and cancer risk. Results will help inform the development of future physical activity/sedentary behaviour related interventions for cancer prevention. Linkage between UK Biobank and cancer registries provides information on cohort members who have been diagnosed with cancers. Using appropriate statistical techniques, we will compare physical activity and sedentary behaviour amongst cohort members who have developed cancer with those who have not developed these cancers, examining questionnaire and accelerometer (measure of physical activity/sedentary behaviour) data. We will also evaluate the associations with relevant biomarkers from the panel of blood/urinary markers and genetic analyses to be undertaken on all cohort members. The full cohort will be included in all analyses to be undertaken.

AIM – Over the next three years, we will develop a versatile AI foundation model that integrates health records, medical images, and genomic data from the UK Biobank. Our goal is to set a new standard in predicting disease risk, forecasting outcomes, and uncovering biological drivers of both common conditions (like cancer) and rare disorders.
BACKGROUND & RATIONALE – Today, clinicians must sift through vast amounts of patient history, imaging scans, lab tests, and genetic information to make treatment decisions. As biomedical data grows, this process becomes more time-consuming and prone to oversight. AI offers the promise of identifying hidden patterns across diverse data types, but existing tools rarely handle records, images, and genomic sequences all at once-especially at the scale of hundreds of thousands of patients. By training our model on such a large, varied population, we expect it to perform accurately for individuals from many genetic and environmental backgrounds.
PUBLIC HEALTH IMPACT – The conditions represented in the UK Biobank affect millions worldwide. Our model will help predict who is at highest risk, estimate likely disease trajectories, and suggest underlying molecular causes. We will share our code and results openly, so other researchers can build on our work. Ultimately, this will lead to earlier testing and diagnosis, more personalized treatment plans, and new insights that drive the discovery of novel therapies.

Age-related hearing loss has been shown to be associated with greater cognitive impairment among older adults; however, the mechanisms underlying this relationship are not completely understood. This project aims to investigate the potential brain mechanisms and pathways linking age-related hearing loss and cognitive decline in older adults, by examining the effect of age-related hearing loss on different measures of brain organization and how these, in turn, relate to cognitive decline. This project will run for a 36-month period, and is expected to increase our understanding of the cognitive and neurobiological correlates of age-related hearing loss, while generating hypotheses for future interventional studies examining the potential benetfits of hearing interventions on brain organization and cognitive performance.

1. Aims:
1.1. To develop a deep learning algorithm for glaucoma using patients’ clinical assessment, epidemiologic data, living environment record and imaging.
1.2. To explore the genetic and metabolomics factors associated with the mechanisms of glaucoma pathogenesis.
1.3 To exploit a multi-modal deep learning strategy to predict glaucoma onset and progression combing clinical, genetic and metabolomics data.

2. Scientific rationale:
Glaucoma is the leading cause of irreversible blindness worldwide.It is a multifactorial disease, whith intraocular pressure (IOP) being the only modifiable risk factor. However, for many glaucoma patients, the disease progresses even with a well-controlled IOP, suggesting that there are other factors contributing to glaucoma. Additionally, the onset and progression of glaucoma is insidious. Patients are aware of the disease until reaching the advanced stages. Thus, it is urgent to recognize the early markers for the disease. Our team is working on a multi-modal deep learning strategy using clinical data, OCT and fundus photography to recognize and grade the severity of glaucomatous visual field damages. Recent GWAS and metabolomics studies have also shown the possible genetic and metabolomic mechanisms of glaucoma. Therefore, combing the clinical, genetic and metabolomics data may help identify novel strategies for predicting.As a result, we will better manage glaucoma beyond controlling eye pressure.

3. Project duration
The project will last 36 months.

4. Public health impact
The UK Biobank dataset comprises a vast number of participants, allowing for robust statistical analyses with increased statistical power to detect associations.Using multi-modal strategies to combing clinical data, genetic and metabolomic information, our research can identify biomarkers and risk factors that allow for earlier detection and intervention, enabling timely treatment to prevent glaucoma progression, and personalized treatment plans based on an individual’s genetic and metabolic profile, optimizing IOP-lowering therapy and reducing side effects. Additionally, integration with other omics data from the UK Biobank could provide a holistic view of glaucoma’s molecular underpinnings for better understanding of the disease.

Early prediction and diagnosis of eye and brain diseases as non-renewable organ-related diseases remain challenging. Although techniques such as retinal imaging have been shown to correlate with pathological processes in the brain, current prediction accuracy and generalizability are limited and further validation is urgently needed. Biological age is an important indicator for assessing the aging process, and the wealth of genetic, proteomic, phenotypic and imaging data in UK Biobank offers the possibility of calculating the aging process in individuals. However, existing studies are still limited in the selection and cross-organizational applicability of biological age metrics.
Therefore, this project intends to integrate multi-omics data (including proteomic, genomic, imaging and clinical information) in UKB, construct an individual aging degree assessment model based on machine learning, and explore its potential application in early prediction of ocular and neurodegenerative diseases. We will combine phenotypic and histological features to identify key biomarkers and systematically analyze their commonalities and specificities in different disease types, providing a theoretical basis for understanding aging mechanisms, disease prediction and individualized intervention.

The main aim of this research is to comprehensively investigate the profound impact of the triple burden of malnutrition (TBM) on the risk of cancer and cardiometabolic diseases (CMDs). Furthermore, the effects of selection bias adjustment for predicting cancer and CMDs risk will be investigated. Accordingly, the research questions and aims are set as follows:
Research Question(s) and Aim(s)
Questions:
* How will the selection bias adjustment affect the prediction of the risk of cancer and CMDs?
* How do the different forms of malnutrition (undernutrition, overnutrition, and micronutrient deficiencies) impact cancer and CMDs?
* What are the mediating effects of the multi-omics (proteomic and metabolomic) factors linking the triple burden of malnutrition with cancer and CMDs?

Aims:
* Assess the effects of selection bias adjustment for the prediction of the risk of cancer and CMDs.
* To compare the impact of different forms of malnutrition on the risk of cancer and CMDs.
* To assess the mediating effects of multi-omics factors linking the TBM with the risk of cancer and CMDs.

This project will develop a novel analysis method and software package able to identify subtypes of disease based on the integration of multiple types of omics data. Many drug candidates fail and many patients receive inappropriate treatment because of our current inability to distinguish between subgroups of patients (respondent vs. non-respondents) and/or subtypes of disease (aggressive vs. non-aggressive). We are developing an approach that will be used both to optimize treatment by separating patients with aggressive disease from those with less aggressive disease, and to increase the success of clinical trials by sub-typing patient groups that are more likely to be respondents from those non-respondents.

Our algorithm will first be used to sub-type patients based on variant data only, then clinical data and omics data will be incorporated to improve the sensitivity and specificity of the sub-typing. UK Biobank is a critical asset for this research because it offers high quality omics and clinical data from a very large population sample. After fine-tuning our algorithm and establishing its effectiveness based on UK Biobank data, we will replicate the research in an independent dataset derived from the US ClinicalTrials.gov database.

We anticipate that this research will significantly increase the effectiveness of clinical trials, and make a contribution to the field consistent with peer-reviewed publication. The proposed work is scheduled for a total of two years.

South Asians have a 2- to 3-fold higher risk of type 2 diabetes and coronary heart disease than White Europeans but may exhibit lower all-cause and cancer mortality (1,2). The reasons for these disparities remain unclear. Omics data-covering metabolomics, proteomics, and whole-genome sequencing-offer a robust lens to uncover the biological mechanisms behind these differences.
Data Sources
1. South Asia Biobank – ~100,000 South Asian participants in the UK, with up to 20 years of follow-up on lifestyle, environment, sociodemographic, clinical markers, genetics, and metabolomics (3).
2. UK Biobank – A large dataset including White Europeans and ~8,000 South Asians, with over 15 years of follow-up.
Research Questions
1. Which lifestyle, genetic, and omics factors best predict cardiovascular disease, type 2 diabetes, cancer, and all-cause mortality etc.?
2. How do these risk profiles differ between South Asians and White Europeans?
3. Can these findings be generalized or adapted to guide disease prevention in other ethnic groups?
Objectives
1. Integrate data from both cohorts to enhance statistical power and capture a broad range of genetic, environmental, and lifestyle factors.
2. Identify and validate key risk factors (e.g., lifestyle, polygenic risk scores, metabolic, and proteomic profiles) associated with disease in each ethnic group.
3. Compare risk factor contributions between South Asians and White Europeans to uncover the main drivers of observed health disparities.
Expected Impact
By clarifying how various risk factors shape disease trajectories, this research will inform targeted prevention strategies and personalized clinical interventions. Although focused on South Asians and White Europeans, insights gained may extend to other populations and guide broader public health measures.
References
1. Gholap N, et al. Primary Care Diabetes. 2011;5(2):45-56.
2. Bhopal RS, et al. PLoS Med. 2018;15(3):e1002515.
3. https://www.sabiobank.org/

With the global population aging, the prevention and control of chronic diseases have become a major challenge. Due to the non-invasive nature of ophthalmic examinations, the eye serves as a crucial window for assessing overall health and detecting systemic diseases. Traditional research has established associations between ocular diseases or specific ocular parameters and multi-omics data.

Recent advancements in artificial intelligence have revolutionized the extraction and analysis of large-scale molecular data. High-dimensional data from various omics disciplines are increasingly accessible, and the integration of different omics data is gaining attention. Multi-level and multi-factor analyses based on genomics, transcriptomics, proteomics, and metabolomics provide a comprehensive perspective on the molecular mechanisms driving disease occurrence and progression. This study aims to elucidate the multifactorial nature of common eye diseases and, by integrating multi-omics with ocular health, identify novel ocular biomarkers for early disease diagnosis, pinpoint potential therapeutic targets, and formulate public health strategies for the prevention and management of chronic eye diseases.

This study aims to deepen the understanding of the complex interactions between multi-omics and ocular health, elucidating the mechanisms linking the two. These insights are essential for developing individualized medical treatments and improving health outcomes across diverse populations.

A) Rationale. Healthy blood vessels are critical to prevent the development of diseases such as heart attack, stroke and dementia, which have an important economical and societal burden in the world. Many risk factors (e.g. obesity, high blood pressure) affect both these diseases of the heart and the brain. In this project, we aim to better understand the different genetic and environmental causes of heart and brain diseases characterized by unhealthy blood vessels.

B) Question and Aims. We want to understand why some individuals suffer from heart and brain diseases, whereas others don’t. In particular, our project will help us identify specific genes which contribute to disease risk (or protection). We will combine this genetic information with other environmental variables (e.g. diet, exercise) and clinical parameters (e.g. cholesterol levels) to develop methods that can predict individuals at higher risk of disease. We will also use the genetic information to better understand what molecules are responsible for disease risk; these molecules could become interesting drug targets.

C) Project duration. This project will last three years.

D) Impact. Heart and brain diseases caused by unhealthy blood vessels are responsible for an important morbidity and mortality. We expect that our project will generate tools to predict disease risk and will help guide the development of more efficacious drug targets.

Research Question & Objectives
Gastrointestinal diseases-colorectal cancer, IBD, NAFLD, peptic ulcer-are rising public-health burdens. Genetic and environmental drivers are known, yet how environmental exposures, psychosocial stressors, and multi-omic profiles interact remains unclear. Using UK Biobank, we will (1) discover novel biomarkers via integrated omics-environment-psychosocial analyses; (2) clarify causal links among exposures, molecular changes, and disease progression; (3) build machine-learning models for early risk prediction and targeted prevention.

Scientific Rationale
Colorectal cancer is a leading cause of cancer death, shaped by genetics, lifestyle, and metabolism. Radiotherapy for pelvic cancers frequently causes chronic radiation-induced intestinal injury. Despite distinct etiology, it shares inflammation and immune dysregulation with other GI diseases, offering a cross-disease model. Circulating proteins, metabolites, and imaging biomarkers illuminate mechanisms. Chronic stress and depressive symptoms amplify gut inflammation and are amplified by immune-metabolic imbalance. Environmental stressors-air pollution, poor diet, sedentary behavior-modulate the gut-brain axis via microbiota, immunity, and neuroendocrine pathways. Most studies examine single omics or isolated exposures, ignoring bidirectional psychosocial interactions. Integrating genomics, proteomics, metabolomics, high-resolution environmental data, mental-health assessments, and accelerometry-derived physical activity will reveal new pathways, refine risk stratification, and guide precision prevention. Advanced causal-inference and machine-learning tools will ensure robust, predictive models ready for translation.

Virtually every cell of our body follows the 24-hr circadian rhythm of a hypothalamic master pacemaker that evolved in the natural light-dark cycle. Decoding this biologic clock culminated in the Nobel Prize in Physiology or Medicine 2017 for the discovery of molecular mechanisms controlling the circadian rhythm. It is now recognized that a strong, unperturbed biologic clock is a hallmark of healthy aging. The introduction of electric light, however, presents unique challenges: today, 20% of the global work force engages in alternate working hours associated with light in unnatural times (eg. night work). Increases in the risk of major chronic disease and mortality have been associated with night work. Further, the ubiquity of light at night implicates potential risk for everyone.
This 5-year project targets the urgency of preventing adverse health consequences of a challenged clock. It does so by aiming to decipher individual risk and related mechanisms 1) using a cutting-edge multi-polygenic score approach; 2) employing transgenerational and deeply phenotyped cohort approaches; and 3) carrying out interventions using genetic risk stratification.
The ultimate aims of this project are to 1) identify the night worker who develops disease, 2) profile mechanisms involved, and 3) optimize effectiveness of interventions to improve sleep and shift work disorder, facilitating immediate implementation of far reaching risk-based prevention strategies/policies, aimed at promoting healthy aging despite clock challenges.

Aims:
Our project aims to uncover the lingering risks of cardiovascular diseases in people who have managed to get their cholesterol and other blood fats to healthy levels. We want to find out why some people still face health issues related to heart diseases even when their lipid levels are under control.
Scientific Rationale:
Cardiovascular diseases are a major cause of death and disability worldwide. High cholesterol and other lipid abnormalities are key risk factors, but even when these are well-managed, there’s still a risk. We suspect this leftover risk could be due to factors like diet, genes, and how the body processes certain substances. New technologies are giving us a chance to look at this in more detail.
Project Duration:
We plan to spend several years on this research to gather comprehensive data from participants, perform detailed analyses of dietary habits, plasma proteomics, metabolites, and polygenic risk scores, and then construct and validate a risk assessment model.
Public Health Impact:
Through this study, we aim to gain a more comprehensive understanding of the residual cardiovascular risk in patients with well-controlled dyslipidemia. By providing more precise risk assessment and management strategies for clinical practice, we hope to ultimately improve patient outcomes and quality of life. This research will contribute to the development of personalized medicine approaches, tailoring interventions to individual patient profiles, and enhancing the effectiveness of preventive measures against cardiovascular diseases.

Sleep is not a commodity but an essential biological process. Good quality and sufficient sleep helps to maintain health, well-being and performance, while poor sleep has been found to put people at risk, both in the short-term (acute effects) and long-term (chronic effects). But what makes people sleep better or worse, later or earlier, more or less, regularly or irregularly? And which of these sleep aspects influence which aspects of health?

In this project, we seek to use the extensive data of the UK Biobank to shed more light on the factors that can influence sleep as well as the possible consequences for health. The factors that we will study are individual factors such as participants’ genetics and chronotype, environmental factors such as season or weather as well as sleep factors such as how and when an individual slept the night before. To assess potential health consequences, we will be investigating the relationship between various sleep features and common diseases that affect large parts of the population e.g. diabetes, atherosclerosis or depression.

We aim to expand on previous findings by deriving new measures of sleep habits and circadian variability, primarily through accelerometer data, investigating the contribution of genetics on new and existing sleep measures through analysis of genotyping or sequence data and explore the long-term effects of sleep and circadian dysfunction on disease risk and progression by incorporating information on genetic predisposition with ongoing healthcare records. We intend to build a more complete picture of how a combination of genetics and environment (and their interaction) contribute to the range of sleep habits (and dysfunction) seen amongst the population.
The project is planned to be long-lived, incorporating new methodology, data sources and expertise as they arise. The initial period will be three years. The insights from our project should aid both basic research and public health. A better understanding of the importance of sleep for healthy living and better knowledge about which factors lead to poor sleep enables better public health campaigns and prioritisation for preventive medicine to improve population health. Insights into the genetics of sleep patterns can help to identify new sleep medicines. Finally, the development of new analysis tools, incorporation of external environmental data and extraction of novel sleep measures will benefit the wider research community and accelerate sleep research.

Research Objective
This study aims to evaluate the associations between Life’s Essential 9 (LE9) and sarcopenia, frailty, and age-related cardiovascular diseases (CVDs) in the UK Biobank (UKB) cohort.
Scientific Rationale
With global aging, sarcopenia, frailty, and CVDs have become interrelated conditions that significantly affect older adults. In 2024, Circulation proposed LE9-a comprehensive framework including four behavioral metrics (diet, physical activity, nicotine exposure, sleep) and five health factors (BMI, blood pressure, blood glucose, blood lipids, and psychological health)-as an updated measure of cardiovascular health (CVH).
Sarcopenia is not only a marker of physical decline but also associated with CVDs. few studies have examined how LE9 components, especially modifiable factors like diet, physical activity, sleep, and mental health, interact with sarcopenia, frailty, and age-related CVDs (e.g., coronary artery disease, arrhythmia, heart failure). It remains unclear whether optimal CVH can offset genetic or environmental risks.
We aim to utilize the entire UK Biobank cohort, including data on socioeconomic status, lifestyle and environmental factors, genomics (such as aging-related clonal hematopoiesis [CHIP], sarcopenia-related whole exome sequencing, whole genome sequencing, and genotyping), biochemical markers, and health outcomes (death/cancer registries, cardiometabolic diseases, and first occurrences). We will also incorporate online follow-up data (24-hour dietary recalls, cognitive function assessments) and imaging data. We will apply genetic risk scores (GRS) and deep learning methods to assess associations between LE9 and aging-related CVD and explore the underlying biological mechanisms within genetic contexts.
Public Health Significance
This study will show LE9’s value in predicting sarcopenia/frailty and CVD, and reveal interactions between CVH and genetic risk. Findings will promote healthy aging.

Genetic variants can be classified into two different groups according to whether they affect the parts of the genome that tell cells how to make proteins or not. The former are called coding mutations, and they can have large effects in an individual. The latter are called non-coding mutations and their effects are usually more modest. So far, we have identified hundreds of non-coding variants associated to higher or lower cancer risk, but we have had limited success in finding coding variants.

We believe that our limited success with coding variants is due to the way we analyze them. So far, we either look at the effect of each coding variant on its own, or we group them according to the protein they affect. However, proteins have different parts with different functions, and the effect of each coding variant likely depends on which precise part of the protein it affects. We have recently developed a new computational tool that uses the information of the different parts of the protein to group the genetic coding variants. We will use this new tool to identify new protein regions that are associated with higher or lower cancer risk.

Finally, we will develop an artificial intelligence model that integrates the effects of both, the coding and non-coding variants of each individual, to predict his or her risk of suffering cancer in the future. If successful, this model could help physicians decide whether a person has a high or low risk of cancer. This is extremely important, as people with high risk should be screened more often, whereas people with low risk might be able to avoid some screens and the discomfort that they might have associated.

Loneliness is an increasingly common experience found in the general population, especially in younger adults and the elderly. The experience of loneliness has detrimental outcomes for mental health as well as physical health, with much research showing links between loneliness and both disease and premature death. The effects of loneliness on health and wellbeing are so severe that its associated morbidity and mortality rates are comparable to that of smoking and obesity.

The proposed research project aims to investigate the associations and possible causes of loneliness, taking into account psychological and social factors, as well as differences in the brain, biology and genetics. We hope that by understanding how loneliness appears in the population, we may be able to gain an insight into how we can reduce experiences of loneliness and possibly even treat loneliness in the future. The project will take place over a period of 3 years.

Recent evidence suggests that genetic variations in the proteins that facilitate drug metabolism can identify which patients benefit from which treatment. This translates to a reduction in side effects and better efficacy of medication, thereby reducing the suffering for the individual patient and quicker recovery. This proposed work will determine which individual patient characteristics, including pharmacogenetics, influence medication response, allowing for the construction of novel models and algorithms for personalized medication prescription. This will include an initial investigation of pharmacogenes, genes that are known to have some impact on medication response. The large datasets that comprise the UK Biobank are a key component to the development of this algorithm and will be used in conjunction with big data from the Finnish Biobank as part of the larger PSY-PGx project (https://www.psy-pgx.org/PSY-PGx). Altogether, we anticipate the project lasting 60 months. Ultimately, this project tackles the expected growth in mental illness in the EU, benefit millions of patients and make for a healthier population leading to cost reductions.

For people at risk of chronic disease, it is often better to start treatment early to prevent or delay the disease altogether, rather than the all-too-common approach of waiting until overt disease develops before intervening. However, even if at high risk of the disease of their lifetime, it is often difficult for younger and middle-aged people without overt disease to access newer therapeutics. This is because treatments only become widely available when shown to be cost-effective, and current health economic models (which underpin these cost-effectiveness analyses) often do not account for the high lifetime risk of chronic disease in some individuals.
Genetic risk scores can now predict the long-term risk of chronic disease from birth, and thus it is possible to identify individuals most likely to benefit from early intervention to prevent disease. In this project, we will use genetic risk scores to identify populations at high lifetime risk of chronic disease, and evaluate the costs and benefits of targeting health interventions to them early in life, before their chronic disease develops.
This will involve first assessing which individuals, by virtue of their genome, are likely to benefit most from certain interventions (for example, some individuals have genetically high cholesterol levels that predispose them to heart disease and are thus likely to benefit most from early cholesterol-lowering treatments). We can then estimate the lifetime risk of chronic disease in these individuals. These estimates then form the basis of our health economic model. We can then use this model to project the costs and benefit of intervening (compared to not intervening).
Unfortunately, due to limited healthcare resources, just proving benefit of an intervention does not guarantee that payers (the government and/or health insurers) will make this intervention available, particularly when the intervention is expensive. Payers usually require evidence of cost-effectiveness to provide access to interventions. Therefore, the key outcome of this project will be to develop a new framework for how new interventions are made available, or how existing interventions targeted to new populations, with the ultimate aim of increasing access to healthcare in a cost-effective manner.

Globally, pancreatic cancer (PaCa) is the 12th most common cancer and the 7th leading cause of cancer death. In the UK, PaCa is the 10th most common cancer, with approximately 10,449 people diagnosed annually (year 2016-2018). Unlike the apparent improving survival rates for common cancers such as breast and colorectal cancer, not much progress in terms of survival rate has been made for PaCa. The 1-year survival rate of PaCa is approximately 28%, and the 5-year survival rate is about 6%. However, Cancer Research UK(CRUK) has reported that 37% of pancreatic cancer patients in the UK can be prevented. Due to the emerging increased incidence and poor prognosis of PaCa, it is crucial to confirm the PaCa factors and target people at risk in the UK population.

Previously published literature has reported many potential PaCa risk factors; however, there were some inconsistent findings. Some probable risk factors include ageing, male gender, African American ethnicity, cigarette smoking, heavy alcohol consumption, increased body mass index (BMI), abdominal obesity, chronic pancreatitis, Diabetes mellitus (DM), hepatitis B and cholecystectomy, family history, germline mutation, ABO blood type and genetic predisposition, etc. Some inconclusive potential risk factors include low physical activity, increased consumption of red/processed meat and dairy products, Vitamin D insufficiency, Helicobacter pylori (H. pylori) infection, long-term use of Proton-pump inhibitors(PPI), and Systemic Lupus Erythematosus(SLE) and anti-diabetic drugs, etc. On the other hand, there is a lack of evidence from a large population-based cohort study. Moreover, there was no comprehensive predictive model to combine the lifestyle-related factors, medical-related factors and genetic disposition in the previous survey.

This study would like to investigate the comprehensive PaCa risk factors through the UK Biobank cohort. The UK Biobank is a large population-based prospective cohort study that collected data from around half a million UK people. The recruiting was carried out from 22 centres to ensure that population coverage was dispersed across the UK. This study aims to explore potential risk factors, including lifestyle-related risk factors, medical-related risk factors and genetic predisposition in the UK population. Furthermore, a comprehensive predictive model will be established based on these risk factors, which may provide a reference for clinicians and researchers for the following pancreatic cancer-related risk stratification, prevention, and early detection studies. The findings could also be used as a reference to target people at risk of pancreatic cancer in the UK community to promote the pancreatic cancer prevention program.

We currently do not know if there is a shared genetic risk for adult-onset obesity and substance use disorders. The aims of this research project are: (1) to uncover any genetic risk factors associated with reward system function, including palatable food intake and substance use, and (2) to use these genetic risk factors to improve the prediction of obesity and substance use disorders in the clinic. By investigating a common genetic signal for the reward system, this research project may improve prediction of other adult-onset complex diseases. The duration of this project will last approximately 18 months.

Complex diseases and traits arise from the interplay of genetic and environmental factors. Understanding these interactions is crucial, yet current statistical methods for genotype-by-environment (GxE) interaction analysis often fail to accurately determine causal directions. This project proposes the development of a novel framework, the Genetic Causality Inference Model (GCIM), which infers causal relationships without assuming a predefined direction. Many existing methods rely on prior knowledge of the causal direction in GxE interactions, and if these assumptions are incorrect, they can lead to misleading conclusions.
The primary objective of this project is to develop GCIM to accurately determine the causal directions of GxE interactions. The model will be validated through simulations with varying scenarios, and its performance will be compared to that of existing GxE methods. The model will also be applied to UK Biobank data, which offers comprehensive phenotypic, genetic, and environmental data essential for robust GxE analyses. This dataset will allow us to analyze the causal directions of complex traits and diseases with rigorous quality control measures and confounding adjustments to ensure the reliability of our findings.
This project seeks to address a common limitation of current GxE methods: the potential for misinterpretation when the true causal direction is unknown. GCIM overcomes this by inferring causal directions without predefined assumptions. Through simulations and application to UK Biobank data, the proposed method will be thoroughly validated and verified. The validated method will enhance GxE analysis in complex diseases. The developed framework will be integrated into user-friendly software and made publicly available. The findings will be disseminated through high-impact publications and scientific platforms, promoting collaboration and contributing to personalized medicine by incorporating individual environmental exposures.

Advanced metabolomics methods have now made it possible to measure thousands of small molecules in body specimen including blood. These molecules are important since they could inform us regarding human health and could in some instances be used as target for prevention or treatment of diseases. Although many studies have investigated the relation of the blood levels of these molecules with diseases, its not clear whether the relation is causal. A third factor might be driving the relation or the blood levels of the metabolite might have been afected by the early stages of the disease. Assessment of the causal effect is normally done in a trial setting where participants receive – in random – medications of other interventions that affect the blood levels of the metabolite. However, specific interventions for these molecules are not known and in many instances such studies are not ethical. An alternative approach is to use genetic information (so called Mendelian Randomisation approach). In this approach, genetic variants that are related to blood levels of the metabolite are studied for their relation with clinical traits and disorders. Given that genetic variants are inheritred in random, any differences in the comparison group should be due to the causal effect of the metabilte.

In another projedct, we have identified genetic variants for a wide range of metabolies and are interested to test their associations with traits and diseases. In this project, we wil use data from UK bio bank to study the association of the genetic variants with a wide range of clinical traits and diseases. This will unravel the clinical importance of the studied metabolites in an agnostic approach and the results could be used either to build or improve risk prediction models or provide novel drug target candidates.

In the entire world, Alzheimer’s disease is the number one cause of morbidity and mortality. The need for primary prevention has been reinforced by a clear decline in dementia prevalence and incidence that has been linked to earlier population-level investments like better vascular health and education. However, only a small number of modifiable risk factors were found, and if they were properly managed, they only reduced the risk of disease by 40%. There are still many unidentified risk factors. In this study, we seek to identify novel environmental factors that can be changed that influence the onset of Alzheimer’s disease and to identify potential targets for interventions. To be more precise, we will 1) evaluate the impact of various environmental risk factors, such as lifestyles, comorbidity, medications, local environmental exposures, etc.; 2) investigate the potential interaction between these different factors; 3) identify people with genetic susceptibility and the way effect of risk factors can be modified by genetic variation; and 4) assess whether lifestyle factors play a role as potential effect modulator. A better understanding of the interplay between different modifiable environmental factors and the development of Alzheimer’s disease is expected to provide new evidence for intervention target and facilitate more efficient prevention strategies and. This project may lead to a more holistic approach to personalized prevention and treatment, accounting for both individual-level factors and community-level environmental factors.

The aim of this study was to systematically screen and validate a wide range of potential risk factors for Parkinson’s disease (PD). Current studies have found that the incidence of PD is closely related to many factors, among which genetic factors play an important role. Large-scale Genome wide association study (GWAS) analysis further confirmed the role of gene variation in PD incidence. However, current studies mainly discuss the relationship between PD and risk factors by setting risk factors, and many risk factors are still ignored or unknown. The phenome-wide association study (PheWAS) are hypothesis-free analyses that examine genetic associations of a wide range of factors with disease. In view of the fact that genetic association can be used as a tool to identify risk factors for PD, this study intends to apply the polygenic risk score (PRS) of PD to represent the risk of PD, which can systematically screen a wide range of established and unknown non-genetic factors associated with PD. On this basis, two-sample Mendelian randomization and brain structures association analysis were used to further evaluate the potential causal relationship between the identified factors and PD and explore the potential biological mechanism of PD-PRS and PD-related factors. During this study, which is expected to be completed within 12 months, a comprehensive and rigorous assessment of the association between a wide range of risk factors and PD will be conducted. The systematic integration of genetic, clinical and neuroimaging information will help to identify risk factors and prevention goals for PD.

Cardiovascular diseases have been known to associate with genetic predisposition, however, the pleiotropic effects of a single genetic variant or group of genetic variants on various disease phenotypes have not been well understood. Phenome-wide association studies (PheWAS) are defined as a statistical approach to test for the association of a genetic variant or accumulative effects of genetic variants and a wide range of phenotypes. PheWAS has been widely used as the large genetic data set has recently become available. PheWAS is particularly useful in situations where we currently have an incomplete understanding of disease mechanisms. Moreover, PheWAS approaches were based on genotypes that are fixed from birth, so it is less affected by environmental confounding factors and reverse causality. This research project aims to investigate the phenotypes that were associated with a polygenic risk score for cardiovascular diseases, and to study whether there is a causal association of cardiovascular disease-polygenic risk score (PRS) and the phenotypes using mendelian randomization (MR). MR is a widely used epidemiological method that utilizes genetic variants to investigate casual association of a risk factor and an outcome. In MR, genetic variants that satisfy all the assumptions are used as an instrumental variable (IV). Assumption of MR includes that genetic variants are associated with risk factors and IV are not directly associated with confounders of the risk factors, nor does it affect the outcome directly. The goal of the research project is to discover the causal risk factors for cardiovascular diseases and to construct a predictive model for new diagnostic tests or treatments. The project duration is about 36 months. Clinical implications include understanding the genetic causes of cardiovascular diseases and applying such information for clinical practice. Public health impacts are concerned with discovering and predicting causative risk factors for cardiovascular diseases that can be a cornerstone of timely clinical management.

To identify changes in the DNA, particularly large deletions and duplications of the genome, that are related to complex diseases and intermediate biomarkers The identification of genetic risk factors that are related to disease will help reveal the underlying biology and thereby expose targets for environmental modification and/or novel drug development. We will use the raw Affymetrix Axiom data (.CEL files) and apply a principal components analysis on these intensities to correct for DNA quality and batch effects. We will then perform a joint calling across all samples to determine potential deletions and duplications using our soon to be open-source Genvisis software package, which will also assess the quality of these calls and filter them down to a set of high quality calls (~1 year to complete). We will then associate the high quality calls with the phenotypes available, and replicate any findings using external data sources (~1 additional year). Full cohort; all samples with high quality intensity data will be analyzed

With the global rise in the elderly population, there has been a notable increase in the occurrence of diseases and mortality rates, with dementia being a major contributor. Dementia not only leads to functional impairments but also reduces life expectancy, making it a critical concern for public health. However, most existing studies on dementia primarily rely on epidemiological analyses, lacking comprehensive pre-clinical brain image features and omics-based analyses that are crucial for understanding the development of early-stage dementia. Additionally, although previous research has explored risk factors associated with dementia, such as diet, physical activity, sleep quality, metabolites, air pollution, and living in greener environments, no systematic studies have been conducted to assess the combined effect of these factors. Meanwhile, the precise mechanism by which these factors influence the risk of dementia, particularly the pre-clinical brain pathological changes, remains largely unclear, making early-stage dementia development elusive. Moreover, while incident multimorbidity, including conditions like obesity, stroke, depression, and sleep disturbances, has been reported to be associated with dementia, validating these associations requires large-scale prospective studies.
Understanding the onset of dementia is of clinical significance as it can aid in prevention and delay the progression of the disease. Therefore, our project aims to (1) examine the independent and joint associations of social, dietary, behavioral, and environmental exposomes with dementia-related brain changes and incidence using prospective cohort studies; (2) uncover the underlying mechanisms of exposomes on dementia risks through omics and genetic analysis; (3) estimate the temporal patterns of multimorbidity leading to dementia transmission, highlighting the relationships and potential mechanisms among exposomes, multimorbidity, and dementia risks. To achieve these objectives, we will conduct a prospective observational analysis, genome-wide association analysis, phenome-wide association analysis, and mendelian randomization analysis. Additionally, a series of secondary analyses will be performed using data from the UK Biobank. Our plan is to begin the analyses as soon as the data become available and aim to complete the project within 36 months, including manuscript preparation for review by authors.
We hope that our study will provide a comprehensive understanding of the associations between exposomes, sleep architectures, aging-related multimorbidity, and pre-clinical brain pathological changes. Additionally, we anticipate identifying novel biological pathways and potential preventive targets for improving dementia prevention. Our project’s objectives align with the goals of the UK Biobank, which is dedicated to enhancing the prevention, diagnosis, and treatment of various serious and life-threatening illnesses, including dementia.

It has been shown through many population studies that Statin usage causes an increased risk of developing type two diabetes. Since Statins have been shown to be safe and effective at lowering the risk of cardiovascular disease, they will remain a widely used first line therapy for many people. Therefore, it is clinically relevant to understand the underlying mechanisms of this risk so both preventive and therapeutic actions can be taken.
To explore this risk, we propose to use polygenic risk scores – which describe someone’s genetic predisposition for acquiring a certain disease. Overall, we want to see if those that have a higher genetic risk for certain diseases like diabetes or high cholesterol, have a higher than expected risk of developing new onset type two diabetes when taking Statins.

We have three main aims in this project, the first is to create a model to predict the risk of new onset type 2 diabetes using polygenic risk scores and other factors such as age, BMI and Statin usage. The second is to use this model to test for greater than expected diabetes risk in those taking Statins who have a high polygenetic risk score. And finally, if an interaction exists, test for interactions between single genetic variants that make up the polygenic risk score and Statin usage.

The results of this study may be able to identify what groups are at greater risk for increased diabetes susceptibility related to Statin usage and by what biological pathways this increased risk operates through. Having knowledge of what groups are at a higher risk may enable preventative measures to be taken to reduce this risk. Additionally, knowing what biological pathways this increased risk may be working through could provide insights into therapeutic actions that may be taken to mitigate risk.

This project will take approximately 2 years to complete.

Our research project aims to uncover new insights into human diseases and improve patient outcomes by identifying specific markers and potential treatment targets using cutting-edge technologies.
Human diseases, such as heart disease, diabetes, cancer, and neurodegenerative disorders, affect the lives of millions of people worldwide. Despite extensive research, many aspects of these diseases remain unclear. By analyzing a large and diverse dataset from the UK Biobank, which includes genetic, proteomic, metabolomic, and multi-modal imaging information as well as multiple lines of disease information, we will gain a deeper understanding of the complex biological mechanisms underlying these diseases. By studying multiple layers of biological information simultaneously, we can identify common molecular patterns across different chronic diseases and uncover specific biomarkers that can predict disease susceptibility, progression, and response to treatment. The estimated duration of the project is 36 months. The knowledge generated by this project will help in developing personalized and targeted interventions, ultimately improving patient care and reducing the burden of chronic diseases on individuals and society as a whole.

As people age, their blood accumulates genetic changes. One type of change, called “CHIP”, strongly predicts whether a person will soon develop disease such as cancer or heart disease. If we could develop a method to detect CHIP in healthy people, it could revolutionise how these common diseases are managed. Unfortunately, current methods to identify CHIP are slow and expensive, and not ready for application to public health.

We believe we have developed a method to address this, and enable the detection of CHIP using standard techniques that are already used in clinical laboratories. We will use the UK Biobank data to refine and test our methods, and establish whether our measure of CHIP predicts future health in the UK Biobank. This project will take around two years, and if successful has the potential to establish a new approach to identify people at high risk of cancer and heart disease in the population, for targeted surveillance and disease management.

his study aims to uncover human genome and examine their roles in the promotion of health and welfare. In this study, we will identify genes associated with available clinical phenotypes and quantitative traits in UK Biobank by using genome-wide PheWAS and evaluate how they influence the studied diseases or quantitative traits. This is the first and fundamental step to understand disease etiology and genetic mechanism. Complex diseases are co-influenced by genes, environmental exposures, and their interaction. This study will also examine how disease genes work with environmental exposures to increase a risk of the studied diseases or quantitative traits. The empirical evidence from this study will provide a better understanding about our genome. The information has a potential clinical impact for developing disease prevention, diagnosis and treatment in precision health that different treatments are applied to different patients according to their unique genomic make up. We plan to execute this project for a 3-year period that includes data clean, data analysis, data interpretation, and results presentation and publications.

Alzheimer?s disease (AD) has reached global epidemic proportions. Therapeutics to prevent, delay and treat AD are urgently needed.

Significant emerging evidence links cholesterol, Aß and AD, and several studies have shown a reduced risk for AD and dementia in populations treated with statins. The ApoE4 allele of the apolipoprotein E gene, is associated with higher cholesterol levels and increased risk for AD. Preliminary results of our own meta-analysis of clinical trial data indicate that the use of statins may delay or slow down the progression of the disease and cognitive decline, to a greater extent in ApoE4 carriers. Despite substantial research and development investment in Alzheimer?s disease, effective therapeutics remain elusive. Our precision medicine approach will generate scientific evidence needed to drive the development of therapies that treat the right person, with the right treatment at the right time. We propose the use of a precision medicine approach, which takes into account people’s individual variations in genes, environment and lifestyle, for analysis of data from the UK BioBank, in order to further examine the effect that the use of statins may have on the onset and cognitive decline of Alzheimer’s disease. Full cohort

Hematopoietic aging represents one of the most critical biological processes impacting global health. Epidemiological studies reveal that approximately 23% of individuals aged 65 or older worldwide exhibit detectable biomarkers of hematopoietic stem cell (HSC) senescence. This population demonstrates a 4.1-fold elevated risk of developing hematologic malignancies-including myelodysplastic syndromes and lymphomas-compared to healthy aging cohorts. According to World Health Organization statistics, complications associated with hematopoietic aging (e.g., thromboembolic disorders, septic shock) account for 18% of mortality causes in elderly populations globally, demonstrating clinical significance comparable to cardiovascular diseases.

Molecular alterations such as leukocyte telomere attrition and epigenetic dysregulation in hematopoietic stem cells (HSCs) may precede overt clinical symptoms by two decades. However, current diagnostic protocols lack the capacity to detect these early warning signals, leaving a critical gap in preventive healthcare strategies. Current diagnostic frameworks relying on conventional blood tests remain limited to detecting late-stage pathological alterations, with insufficient resolution to identify early aging biomarkers.

Our aim is to leverage the UK Biobank database to integrate whole-genome sequencing, complete blood count parameters, plasma proteomic profiles, metabolomic profiles, leukocyte telomere length measurements, and detailed clinical data. By applying advanced machine learning techniques, we aim to establish a multimodal hematopoietic aging prediction system. Based on this model, we will also analyze how lifestyle factors influence hematopoietic aging trajectories and the progression of hematologic disorders. This model could help in accurating diagnosis of individual hematopoietic aging predisposition and predict the onset and progression of blood-related diseases at the personalized level.

Cancer and chronic diseases, such as diabetes and cardiovascular conditions, represent major global health burdens, with complex biological interconnections and often shared molecular mechanisms, including genetic mutations, metabolic imbalances, and abnormal protein expression. Biomarkers play a critical role in early diagnosis, prognosis, and therapeutic monitoring, as they reflect physiological states and disease characteristics throughout the onset and progression of these conditions. However, many studies have been limited to observational designs, lacking the prospective research needed to fully realise the clinical potential of biomarkers in diagnosis, prognosis, prediction, and monitoring. Long-term follow-up and large-scale studies are crucial to evaluate the effectiveness of biomarkers in clinical practice, which would substantially enhance their applicability. This prospective, multicentre clinical study aims to establish a multi-omics data analysis platform to systematically identify and validate clinically valuable biomarkers. This approach seeks to improve the accuracy of early diagnosis, predict disease progression, support personalised treatments, and develop predictive and diagnostic models, thus providing a robust scientific basis for precision medicine.

The aim of our project (estimated duration: 2 years) is to investigate the effect of lifestyle on breast cancer risk and prognosis, and to evaluate the efficacy of separated or combined lifestyle interventions on the incidence and prognosis of female breast cancer and further explore the possible mechanisms more and more attention has been paid to the impact of daily lifestyle on the incidence and prognosis of breast cancer. Breast cancer is a complex disease resulting from a combination of genetic and environmental factors. Daily lifestyles were reported to be related to cancer risk and prognosis, including physical activity, electronic device use, sleep, smoking, diet habit, alcohol intake and sexual factors et al. Although some studies have explored and speculated on this issue, these researches are not in-depth enough and the results are controversial. Therefore, solid and thorough population researches of this issue are needed. We plan to discuss the lifestyle risk factors effect on breast cancer risk and prognosis separately according to the clinical characteristics, such as menstrual history, age, reproductive history, etc. Lifestyle risk factors can be stratified by severity, such as smoking and alcohol intake, time spent on electronics, and frequency of physical activity The study will provide lifestyle guidance for healthy people and breast cancer patients, provide strategies to prevent breast cancer and improve prognosis, reduce the incidence and mortality of breast cancer, and reduce the medical burden of social funds.

This project aims to investigate how clinical characteristics, blood proteomics, and metabolomics influence the incidence and prognosis of urological cancers (kidney, prostate, and bladder) in the UK Biobank cohort. We will address three key research questions: 1.Which clinical factors (e.g., age, BMI, lifestyle variables) are significantly associated with urological cancer incidence and survival? 2.How do proteomics and metabolomics markers improve risk stratification and prediction of disease outcomes? 3.Can machine learning models leveraging these multi-dimensional data offer clinically valuable diagnostic and prognostic tools?Objectives:Conduct Cox proportional hazards analyses to identify associations between baseline clinical features and cancer outcomes.Evaluate the predictive utility of selected protein and metabolic biomarkers using both traditional and machine learning approaches.Develop and validate diagnostic/prognostic models, comparing their accuracy to established clinical risk scores.Scientific Rationale:Although multiple studies have examined risk factors for urological cancers, few have integrated comprehensive proteomic and metabolomic data at a large scale. By leveraging the breadth and depth of UK Biobank data, our research will clarify the interplay between traditional risk factors and molecular biomarkers. The resulting findings may guide personalized risk assessment, improve early detection, and inform targeted prevention strategies for urological malignancies.

This project will provide a general method for genome data analysis that can be potentially apply to any specific disease. We focus the application of this method on an investigation into the role of genetic polymorphisms in bladder cancer. This will advance our knowledge of bladder cancer and may help to improve diagnosis, prevention, and treatment for bladder cancer.

Full cohort, with the possibility of adding data from the future web-based questionnaire on pain.

Aims: Our research project aims to investigate the association between childhood trauma experience (measured by Childhood Trauma Screener) and bone health markers in adults, including bone-related serum markers and bone mineral density, through retrospective analyzing a large database cohort, thus to infer that whether childhood trauma experience has an association with bone health. We also aim to explore further which Childhood Trauma Screener subtype is associated with bone health.

Scientific rationale: Our previous research has found that the experience of childhood trauma impact bone health in adults, including reduced bone mineral density and dysregulated serum bone markers through animal and human studies. Now we try to take the research a step further, by analyzing the data from the UK Biobank database, we can explore the association between childhood trauma and bone health in a large cohort, updates our understanding of osteoporosis etiology, and open new avenues for personalized prevention and intervention strategies.

Project duration: The estimated duration of this project is 36 months.

Public health impact: The results of this study allow for additional beneficial outcomes for public health. By identifying potential associations with childhood trauma, the results of this study have the potential to provide early prevention strategies and increase awareness among healthcare practitioners, policymakers, and the public. Then will ultimately contribute to reducing the population’s childhood trauma burden.

We aim to look at the movement data in UK Biobank data using a computer program that has been designed to spot patterns in large volumes of data. We will anonymously group people by medical history, age and gender (amongst other criteria), and observe for similarities within these groups. We will then tell the computer program what some of these represent, and it will learn to recognise changes from these patterns in the future – for example, the tremor observed in Parkinsonism and the limp seen in Osteoarthritis of the hip. It is possible that the program will discover similarities that have not been previously noticed by people. This might mean that we are able to discover new patterns of movement, that can be used to help diagnose people with illnesses in the future.

We believe that this project will take 1-2 years to complete fully, as the development of the computer program will be dependent on the data that we receive (and so it is very difficult to predict the duration).

This project presents the potential to help diagnose illness on a large scale by using the technological advancements (for instance, smart watches) that people already use. The opportunity to find new movement patterns is also an exciting one, and may be something that could help people even without access to wearable technology.

CVD is the second leading cause of death in Canada. Every year, approximately 70,000 Canadians die from a CVD-related illness and many more suffer life-threatening CVD events such as myocardial infarctions and strokes.
The research aims of the proposed study are:
Primary objective: to build, validate and optimize a multimodal architecture using deep learning algorithms to predict cardiovascular outcomes based on imaging data (MRI and ultrasound), ECG and patient clinical information.
Secondary objective: To determine new clinical prognostic parameters of cardiovascular outcomes based on the multimodal architecture, and to determine the impact of the population- and sex- differences in cardiovascular outcomes.
Most of the existing methods have been focusing on analyzing 2D images analysis rather than 3D images that are commonly prescribed to diagnose CVD. Moreover, most of the existing methods solely focus on analyzing images but neglect other information such as health records that can also be crucial in diagnosing CVD.
Our study aims to create a deep learning multimodal architecture to predict CVD outcomes by integrating complex imaging segmentation analysis for 3D images and multimodal data input such as clinical notes and ECG records. Our proposed architecture is promising to outperform the current standards given that deep learning methods are recognized to identify non-linear relationships in high-dimensional data.
The proposed study is expected to accelerate the process of 3D image analysis, to improve the accuracy of CVD diagnosis by incorporating 3D images analysis results with other clinical measurements, and ultimately to facilitate practitioners in better clinical decision-making.
The estimated duration of the study is 36 months: 6 months for data processing, 18 months for the primary objective and 12 months for the secondary objective.

Frailty is characterised by weakening of body’s normal functioning and can cause unwanted health problems such as hospitalisation, falls, sensitivity to unwanted effects of medications and death. Frailty can significantly increase the health-care costs and is a global health concern, and its incidence is expected to rise due to the rapid increase in ageing population.
Depression is one of the most common mental health illnesses worldwide and can have significant negative effect on the quality of life of patients and their care givers and cause death mainly due to suicide.
Depression is treated mostly in primary care settings such as GP surgeries in the UK. A group of medications called Selective Serotonin Reuptake Inhibitors (SSRIs) are the first line antidepressant drugs for treatment of depression in adults in the UK. They can cause unwanted effects such as falls, irregular heartbeat, anxiety, weakness of bones, nausea and feeling sick, bleeding, and sexual problems. They can also interact with other medications.
Frail patients generally have higher possibility of being diagnosed with depression. Frail patients may respond less favourably to the antidepressants and are more sensitive to their unwanted effects. Many frail patients have other illnesses and therefore are often on numerous medications which can interact with antidepressants and induce further risks and complications.
There is a significant gap of knowledge in relation to the role of SSRIs in causing or worsening frailty. The current treatment guidelines for treatment of depression do not provide sufficient advice about risks and benefits of using SSRIs for treatment of depression in frail patients.
This research investigates whether the use of SSRIs for treatment of depression can cause or worsen frailty in primary care patients. It also investigates the risks and benefits of using SSRIs as the first line antidepressants for treatment of depression in frail primary care patients and possible links between genetic and patient characteristics and responding to the SSRIs or having unwanted effects.
This project aims to develop a screening tool to help clinicians to prescribe SSRIs for frail patients when are expected to be effective and less likely to cause unwanted side effects to reduce the harm to patients and ensure effective treatment and is expected to last 36 months.

The prevalence of hypertension correlates with latitude, with BP rising with distance from the equator. Vitamin D is made in sun exposed skin, and serum levels correlate inversely with hypertension and ischaemic heart disease(IHD) prevalence. However, oral vitamin D supplements have no effect on BP or IHD. We hypothesise that sunlight has beneficial effects on cardiovascular disease and overall mortality, independently of vitamin D. We have demonstrated large stores of nitrogen oxides in the skin, released to the circulation on UVA irradiation of the skin, with a fall in BP.
Although sunlight exposure is risk factor for skin cancer there is no evidence that it causes a rise in all-cause mortality. epidemiological data suggest it may cause a reduction in deaths from hypertension related diseases, in particular cardiovascular and cerebrovascular disease. As deaths from these are almost 100 times more frequent than from skin cancer, the benefits of sunlight may exceed the risks. The UK Biobank dataset includes questions on sun exposure and latitude of residence and allow a prospective study on overall risk-benefit ratio of sunlight exposure. This meets the aspirations of the UK Biobank to improve prevention and treatment of disease, in particular cardiovascular and stroke.
We will seek access to the full dataset, looking at data on sunlight exposure and variables known to affect cardiovascular disease and stroke. Our primary outcome measure will be all-cause mortality and we will derive an odds-ratio for the effect of sunlight on this. We will also look for rates of skin cancer. We will correct for confounding factors that may link with different sun exposure patterns. We require access to data only (i.e. no samples) for the full cohort, with data on vitamin D serum levels when available

Livers have the second longest transplant waitlist in the US next to kidneys. Alcohol and obesity are known causes of fat build up in the liver. This fat build up leads to scarring. 1 in 4 people in the US and Europe have livers that are over 5% fat and experience no symptoms. The only good options for quantifying liver fat percentage are an invasive liver biopsy or an expensive MRI, both of which require significant suspicion of liver damage or other problems to justify. A tighter feedback loop would allow healthy minded individuals to correct deleterious behavior before reversible liver fat becomes irreversible liver scarring.

We aim to combine the latest advances in AI with routine blood and urine measurements to create a tool with which the healthy man and woman can better monitor their liver health. The current medically accepted fatty liver calculator uses 2 blood measurements in conjunction with body weight and height. It was trained on data collected from 496 people. The UK Biobank dataset contains data from over half a million people, and tons of blood and urine measurements. Our scientific rationale is that we can use this enormous dataset to train a more reliable fatty liver estimator.

When granted access to the dataset, we expect the project to be completed in under 3 months. If the trained estimator is reliable, we plan to release it to the public as an app. The public health impact of this work will be to allow better monitoring of liver health with data collected from routine blood and urine lab measurements.

This project aims to determine whether there is a difference between colorectal cancer patients one year pre-diagnosis and people without cancer diagnoses in their dietary intakes based on healthy eating recommendations. There is evidence highlighted that adopting a healthy diet may help in cancer prevention and improve survivorship. Furthermore, higher rates of mortality and morbidity noted in cancer survivors compared to people who have never had cancer, have motivated people living with and beyond cancer to change their lifestyle to a healthier one. Most studies assessed dietary intake before and after diagnosis among breast cancer patients, while very few studies evaluated the same for colorectal cancer patients. Therefore, the present study aims to evaluate dietary intake before and after diagnosis among colorectal cancer patients in the UK. The study’s findings might help patients understand how a healthier diet can facilitate cancer prevention and survivorship as diagnoses become more prevalent. In addition, the study will determine whether colorectal cancer patients tend to change their eating habits after diagnosis and whether this creates an opportunity for overall lifestyle and health improvement. This project will take 36 months from bio bank application submission until disseminate results.

Aims: Our overarching aim is to explore eye-brain function relationships. Glaucoma is a condition which there is gradual damage to the optic nerve, or the nerve of sight. While the optic nerve is located at the back of the eye it is actually an extension of the brain or central nervous system. It is recognised that one of the processes that contributes to glaucoma progression is neuroinflammation, or inflammation of the nervous system. The focus of this research is to identify and evaluate biological, behavioural, and environmental factors in the aetiology of glaucoma and related neurodegenerative disorders in which neuroinflammation plays a significant role.

Scientific Rationale: Glaucoma is regarded as a neurodegenerative disorder. However, the evidence for this hypothesis is controversial, as it is based on generally smaller studies that have reported variable results. Using the powerful and detailed database of UK Biobank, we will robustly evaluate multiple biological, behavioural, and environmental factors implicated in neurodegenerative pathways to determine their causal relevance to glaucoma and related neurodegenerative conditions, thereby identifying modifiable risk factors for glaucoma. In addition, we will develop risk scores which incorporate environmental, biological, physical and genetic variables, thereby advancing risk prediction of glaucoma.

Project duration: 3 years

Potential Public Health Impact: Glaucoma is a leading cause of blindness. Presently, there are no significant modifiable risk factors other than intraocular pressure. This investigation will identify modifiable environmental, physical, biological and genetic biomarkers in relationship to glaucoma. Furthermore, it will explore the association between glaucoma and neurodegenerative pathways which will allow a deeper understanding of the underlying mechanisms that impact the disease process

Aims,
This project aims to improve our understanding of the pathways to subjective wellbeing versus ill-being, taking into account key variables at multiple levels of scale, spanning the individual, community and environment.
Scientific rationale,
Prior research has identified a number of key variables influencing human wellbeing, ranging in scale from biopsychsocial variables (e.g. the beat-to-beat variability in heart rate, cognitive flexibility and social connectedness) to variables relating to sociocontextual and ecological impacts (e.g. social capital, and green space availability etc). Research has seldom accounted for such complexities and potential interactions stemming from the interconnectedness of multiple systemic factors influencing the complex construct of wellbeing and illbeing. This work will therefore conduct structural equation modelling and path analysis to explore more sophisticated model of wellbeing and examine its relevance to related yet distinct construct of illbeing, taking into account the many mediating and moderating factors of key wellbeing markers.
Project duration,
2 Years.
Public health impact,
We will develop a contemporary evidence-based model that reflects the complex construct of individual wellbeing. By incorporating ecological, social, psychological, and physiological influencing factors, this work will facilitate a transdisciplinary approach to studying public health relevant factors of ‘wellbeing’ versus ‘ill-being’. In doing so, we hope to aid a conceptual shift within wellbeing research that will open up avenues for interventions promoting positive change at individual, social, and ecological levels.

In recent years there is accumulating evidence that genetics and lifestyle habits interplay in complex ways to affect clinical conditions. We intend to investigate the relationship between genetic factors, dietary habits, anthropometric traits and several clinical outcomes, with the aim of proposing personalized nutrition recommendations. Our main goal is to minimize the risk of obesity and other diet-induced disease via individual-specific nutrition recommendations. The proposed research aims to improve the prevention, diagnosis and treatment of illness and the promotion of health by providing personalized nutrition recommendations, in alignment with the UK biobank?s stated purpose. We intend to identify genetic mechanisms that, in combination with the dietary habits of an individual, can be used to predict clinical outcomes. These findings can be used to provide personalized nutrition recommendations to minimize risk for adverse outcomes. We intend to use all individuals for which both genetic and dietary habits data is available to obtain as much power as possible to achieve significant results (approximately 500,000 individuals).

Our primary aim is to discover the sex-specific genetic variants, regardless of the way they are associated with a certain disease or trait. By setting this aim, we seek to answer the question that 1) what are the genetic variants that have sex-specific effects on a certain disease or traits. And 2), are those variants impacting the disease or traits through their additive effect, other genetic models, or their interactions with sex.

Many diseases, including most of the autoimmune related diseases, show strong disproportionality between genders. Previous researches often focus either the effect of the variant itself, or on interaction of variant with sex. However, there is no method to jointly estimate both effects and discover the sex-specific genetic architecture therefrom.

To achieve the above aim, we will propose a framework to 1) identify sex-specific genetic variants through statistical tests on the joint effect of additive model of the genetic variants, and their interaction with sex; and 2), further distinguish and categorize the types of effecting patterns of the variants, in order to further unravel the biological pathways and mechanisms involving those variants.

By addressing the above questions, discovering the sex-specific genetic architectures of the traits/diseases that shows strong variance in incident rates between sexes, we hope to shed light on finding the genetic causations on those diseases. And furthermore, by categorizing them by patterns of their sex-specific effects on the traits/diseases, we seek to deepen the understanding of the pathways and mechanisms of those diseases.

We are living longer. Consequently, with age we are likely to accumulate multiple chronic diseases like high blood pressure, Type 2 diabetes and heart diseases at the same time. Scientifically, this is referred as multimorbidity. At the moment if affects 27% of all adults in the UK however this is set to increase significantly in the future. This is going to be one of the biggest challenges not only National Health Service but for society as a whole.
We know that as we get older our health declines, however we also know that several modifiable risk factors such as lifestyle factors such as smoking, diet, alcohol intake, how much we move as well as things like pollution also affect our health. Time and time science has shown that people who move less, smoke or have an excessive intake of alcohol and who eat an unhealthy diet are at a higher risk of developing several chronic diseases like heart disease and cancer. New research shows that exposure to unhealthy levels of air pollution adds to the risk of these chronic diseases. Other research also shows that there are non-medical factors such as the level of education, social class, income and the neighbourhood you live in also play their role in our health. However, there is little research that has looked at the impact of all these factors on health.
Each of these factors (sociodemographic, psychosocial, lifestyle and environmental factor) all play a role in the development of multiple chronic conditions. However, what we don’t yet know is how these risk factors which coexist within the same individual interact together. Studying this specifically is a challenge and as a result standard statistical techniques are of limited use in this regard.

Using the UK Biobank Cohort we aim to try understand the complex relationship between the exposure to multiple health risks and the development of multiple chronic diseases. The expected results of this study will help to identify which groups may be more at risk of developing multiple diseases in the future and will require much more targeted public intervention.

We propose to identify causal protein and genetic determinants of hemorrhagic transformation (HT) in ischemic stroke patients through two-sample or Bi-directional Mendelian randomization (MR) using plasma pQTLs and large!scale GWAS summary statistics from MEGASTROKE and UK Biobank. We aim to develop and validate a high performance machine-learning (ML) based risk prediction model for HT by integrating demographic, clinical, genomic, and proteomic features using algorithms like XGBoost. We will interpret model predictions with SHapley Additive exPlanations (SHAP) to elucidate each feature’s contribution and nonlinear relationships with HT risk. Our objectives include assessing discrimination (AUC) and calibration (calibration curves, decision curve analysis) through nested cross!validation and external validation cohorts. We will perform colocalization analyses to confirm shared causal signals between pQTLs and stroke outcomes. Model hyperparameters will be optimized via grid search to balance performance and generalizability.
Finally, we will generate SHAP-derived risk scores and share all analytic code and derived variables within six months of publication to enhance UK Biobank resource utility.
Last, we need samples data to identify potential biomarkers(including gene, proteins, clinicasl data and samples) for predict HT in early stage by using ML methods.

Polycystic Ovarian Syndrome (PCOS) is a common hormonal condition affecting approximately 5-20% of women worldwide. The condition results in the ovaries producing excessive male hormones. This hormonal imbalance can result in irregular or absent menstrual periods, difficulty becoming pregnant and mood disturbances. The condition also increases the chance of developing cardiovascular disease and diabetes even after menopause.

The exact cause of PCOS and how it increases the chance of developing cardiovascular disease and diabetes is not known. Research shows that there may be a genetic link and that PCOS can run in families. There also appears to be a connection with obesity, although the details of how this occurs is not clear.

Recent research has identified long-term inflammation as a possible link between PCOS and insulin resistance. Insulin is a hormone that allows cells in the body to take up glucose, so women with PCOS often develop a state in which their cells cannot use insulin. As a result, the pancreas can produce more insulin to make up for this and over time can lead to diabetes. Women with PCOS also develop abnormalities and inflammation in their fat tissue, which is thought to contribute to diabetes and cardiovascular risk.

Inflammation often results in involvement of the body’s immune system, which include white blood cells. There is not enough current information about how the inflammation that occurs in PCOS affects white blood cell numbers. This study therefore aims to answer this question by using the UK BioBank data which includes 273,469 women, of whom 643 have self-reported as having PCOS. The project duration is expected to be three to six months, but may be longer as more data becomes available. Answering this research question has the potential to better understand PCOS and how it increases the risk of cardiovascular disease and diabetes. With this greater understanding, novel treatment options can be developed to reduce the morbidity associated with PCOS.

The main goal of the research is the systematic integration of genetic and brain activation data to explain the underpinnings of economic and social choice, and use the results to understand neurocognitive and mental health outcomes. We develop a novel approach integrating functional neuro-imaging and the genetic data using both Bayesian and classical statistical methods.

We plan to build a Hierarchical Bayesian model and use other classical statistic method of decision making. Analysis of UK Biobank data will be cross-validated with independent samples available at the University of Minnesota, relevant for the study of schizophrenia and addiction. The research will address health-related questions in the public interest both directly and indirectly.

Directly, through the application of the study to causal factors of schizophrenia and addiction (both drug/alcohol and behavioral).

Indirectly because it will improve the understanding of the impact of cognitive functions on important life outcomes, which eventually affect health conditions. The specific aim of this study is to examine association signals from SNPs, examining how much of the variation is captured by examining all SNPs simultaneously (using Polygenic Risk Scores), and look at the extent to which SNPs that predict variation in two joint sets of phenotypes (in our case, patterns of brain activation and choice).

For this purpose we will use statistical techniques (Gene Set Enrichment Analysis, Group Lasso) testing the hypotheses that specific biological pathways affect jointly the choices and brain activity implementing it. We will integrate several data sets for cross-validation. The research requires data on the full cohort of subjects available. In particular we need the full genotype data and the raw imaging data. We are aware that this will require large data storage, so a clarification is needed. In our team the effort is aimed at integrating the analysis of the UK Biobank data and the Human Connectome Project (http://www.humanconnectome.org/). For this we need in addition to functional and T1 data the T2 data. We also like to be sure that the geometric B0 correction is done properly.

Children with LCA are born blind.  Even though the disease is rare, it has significant long term healthcare implications. A number of different genes have been implicated in the pathogenesis of LCA. Alleles are different forms of the same gene located on the same place on the chromosome. Comparing the frequency of different alleles in the causation of LCA will give a fair idea as to the mutations responsible. This will help in better screening techniques that can be used in the population, to reduce the incidence and prevalence of the disease, thus resulting in reduced blindness in the overall population. Reduced cases of LCA will mean easing the dependency on the healthcare system resources, even if by very little.
We expect the project to last less than 6months, since it will be done insilico, with a simulated computer program.

Critical illness leading to intensive care means suffering and high risk of death for the individual, and costs to society. There are several reasons why patients end up in intensive care units (IVA), but common to them is that they develop failure in vital organs. Sepsis, COVID-19 infection, and surgical trauma are common causes of failure of vital organs. During the past year, these conditions have accounted for a very large proportion of all admissions to ICU in Sweden.
We have previously shown that inherited a.k.a. genetic factors are important in how ill a person gets from the same infection or surgical procedure. The planned study can help us understand which genetic factors are important risk factors for getting severely ill and recovering thereafter. Moreover, these studies can help us understand some of the mechanisms leading to severe illness, increasing the chance of finding treatments to prevent or cure these conditions.

Aims:
Our aim is to explore the association between influencing factors (including biochemical indices, plasma metabolites, lifestyle, environmental factors and genetic factors) and the incidence and progression of cerebrovascular and neurodegenerative diseases, and investigate the association of influencing factors and the images of brain diseases.

Scientific rationale:
Cerebrovascular and neurodegenerative diseases are important public health issues worldwide, causing serious morbidity, mortality and economic burden. Despite extensive previous research efforts, the influencing factors of these diseases remain incompletely understood. We propose a study to identify modifiable and non-modifiable influencing factors for cerebrovascular and neurodegenerative disease using the UK Biobank data. By identifying these factors, we hope to improve our understanding of the pathogenesis of cerebrovascular and neurodegenerative disease, and contribute to the development of more effective prevention and treatment strategies.

Project duration:
This project is expected to last for 36 months.

Public health impact:
By identifying key contributing factors, public health policies can be developed to help mitigate the incidence, ultimately improving population health outcomes and reducing the economic burden associated with cerebrovascular and neurodegenerative diseases.

Neuroimmune diseases, including multiple sclerosis (MS), neuromyelitis optica spectrum disorders (NMOSD), anti-myelin oligodendrocyte glycoprotein-IgG (MOGAD), Guillain-Barre syndrome (GBS) etc., predominantly occur in young and middle-aged individuals with a high rate of disability. The pathogenesis of these diseases remains unclear, and the treatment outcomes are often unsatisfactory. Therefore, exploring the prevention and treatment of neuroimmune diseases is of vital importance. In this study, we will identify risk factors and protective factors influencing disease onset and prognosis from multiple perspectives based on large-sample population data.
In our research, first we will collect baseline data from the population (including general demographic characteristics, medical history, and laboratory data) as well as outcome events (the onset and prognosis of autoimmune diseases of the nervous system). The population will be grouped, and various statistical methods (such as logistic regression, Cox regression and so on) will be employed to compare the differences in various factors between groups, with the aim of identifying potential protective and risk factors that influence the onset and prognosis of neuroimmune diseases. We also intend to compare factors related to neuroimmune disorders with neurodegenerative diseases, as well as anxiety, depression, and other psychosomatic disorders, which will be more conducive to identifying the signature indicators of neuroimmune that we are concerned about and to finding common factors that affect neurological diseases. Furthermore, we will explore the relationship between various factors and autoimmune diseases of the nervous system at the genetic level.

Cardiovascular and cerebrovascular diseases are the primary diseases that threaten human life and health. At present, some related risk factors have been identified and treatment methods have been applied. However, the prognosis is still poor. Therefore, the purpose of our study is to find the risk factors of diseases from multiple perspectives based on the data of a large sample population, including general demographic characteristics, disease history, medication history, blood biochemical indexes and so on. On the basis of traditional risk factors, we can find more connections, such as the association among cardiovascular and cerebrovascular diseases. Therefore, this study aims to guide the precise management of cardiovascular and cerebrovascular diseases, find new risk factors, explore new targets for prevention and treatment, improve prognosis, reduce disease burden, and contribute to human health.

Cognitive impairment (reduced brain functioning) is more common in people with kidney disease than in those without it. We know that people who have cognitive impairment do not engage well with the healthcare system, have a poorer quality of life and are more likely to die early.

Being physically active is known to improve brain function in the general population. Research has shown some benefits of physical exercise in people with severe kidney problems who are on dialysis treatment. However, this has not been studied in detail in those with less severe forms of kidney disease. By studying the effects of physical activity on brain function in people with kidney disease, we may find a new treatment to help to prevent and slow down cognitive impairment in those with kidney disease.

In this study we aim to;

1. Find out how common cognitive impairment is in those with kidney disease in the UK Biobank study population.
2. Find out if there is a relationship between physical activity level and cognitive function in those with kidney disease.
3. Find out if people with lower physical activity levels and kidney problems are more likely to develop cognitive impairment and die than those who have higher activity levels.
4. Find out if we can see where and what type of damage in the brain occurs in these people, by looking at MRI brain scans, and assess relationship with physical activity. This may also help us to better understand what processes are involved in this damage.
5. Find out what other factors might be involved, such as risk factors for heart disease and inflammation (body’s response to infection or injury), in cognitive impairment in people with kidney disease.

Dizziness is a common complaint among patients who visit primary care physicians, neurologists, and otolaryngologists. Some types of dizziness have been found to be familial, which distinguishes them from other causes of dizziness. Familial dizziness has been observed in patients with isolated recurrent attacks of dizziness, hereditary deafness syndromes, and in patients with neurological disorders. While research in these areas has advanced our understanding of dizziness, much remains to be elucidated about its underlying causes. Through analyzing the genome data of the UK Biobank, we aim to identify genetic mutations that may contribute to development of diseases that can cause dizziness. By identifying genetic factors associated with dizziness, researchers can gain insights into the underlying mechanisms of the disorder and potentially develop more targeted and effective treatments. Additionally, identifying individuals at a higher risk for certain types of dizziness would allow earlier intervention and prevention based on their genetic profile.

Skin diseases are highly prevalent, often chronic, and frequently co-occur with systemic conditions such as cardiovascular disease, inflammatory bowel disease, and various malignancies. Approximately 30% of patients with skin disorders present at least one comorbidity, and their risk of cancer-related comorbidities is estimated to be two to five times higher than the general population. However, the underlying mechanisms remain poorly understood. This study aims to address three key questions: (1) What modifiable and genetic factors are associated with the onset of skin diseases and related comorbidities? (2) How do genetic predispositions interact with environmental triggers in influencing disease progression? (3) What shared biological mechanisms underlie these comorbidities?

To answer these questions, we will leverage the UK Biobank’s large-scale, multidimensional dataset-encompassing genetic, environmental, lifestyle, and clinical information-to identify key risk factors and infer causal relationships. We will apply advanced epidemiological and statistical methods, including Cox regression, GWAS, Mendelian randomization, and machine learning, to explore associations, interactions, and disease trajectories. This integrated approach will help uncover the complex interplay between genes and environment in skin disease etiology and comorbidity development. The findings are expected to provide a robust scientific foundation for early prevention, precise risk stratification, and the development of personalized intervention strategies, thereby improving patient outcomes and supporting public health efforts.

Existing studies show that women with PCOS have a significantly higher prevalence of sleep disorders compared to healthy controls, including obstructive sleep apnea, hypersomnia, and insomnia. Subclinical sleep issues such as poor sleep quality, difficulty falling asleep, fatigue, and nighttime awakenings are also common in PCOS. These sleep problems are linked to metabolic abnormalities like insulin resistance, hyperandrogenism, and obesity. However, the causal relationship between sleep disorders and PCOS onset remains unclear, and longitudinal studies on the impact of sleep disorders on PCOS-related comorbidities are lacking.
This study aims to fill this gap through longitudinal and mediation analyses, over a 36-month period. The primary goals are to explore the interactions between sleep patterns, psychological factors, clock gene polymorphisms, environmental exposures, and PCOS-related reproductive, metabolic, and psychological phenotypes. We will also investigate the causal relationship between sleep patterns and PCOS onset and the effects of sleep disorders on comorbidities. Findings will provide evidence for early prevention and personalized interventions for PCOS.
The UK Biobank cohort includes individuals aged 40-69 (2006-2010), many of whom are postmenopausal. Our research will focus on individuals diagnosed with PCOS before menopause, using clinical data (reproductive history, hormone levels) to identify potential PCOS cases. We will examine long-term health outcomes, such as metabolic and cardiovascular health, using longitudinal UK Biobank data, and have reviewed relevant research to identify suitable cases for the study.

Although DCIS is considered a pre-invasive condition, it can progress to invasive breast cancer or death over a number of years. It is unclear which women with DCIS will develop invasive cancer within their lifetime and over what period of time. This project will investigate the risk factors for developing DCIS and its subsequent progression to invasive cancer and aim to quantify a woman?s risk of disease progression based on her lifestyle, dietary, socioeconomic, medical and genetic factors in order to help clinicians tailor diagnosis/screening, follow-up and treatment options according to individual patient risk-profiles. The objective of the UK Biobank is to undertake research in improving the prevention, diagnosis and treatment of a wide range of serious illnesses including breast cancer. By investigating the risk factors for developing DCIS or having a poorer outcome after DCIS treatment, it may be possible to quantify a woman?s risk based on her individual risk factors and so tailor prevention, diagnosis, and treatment options accordingly. Hence the aim of this project is in synergy with that of the UK Biobank and could help to provide innovative new directions to improve healthcare for DCIS patients. Of the women within the UK Biobank diagnosed with DCIS, analyses will be conducted of patient, medical, socioeconomic, dietary, lifestyle and genetic factors to investigate possible risk factors for DCIS when compared against age matched controls without DCIS or invasive breast cancer and with those women who were diagnosed with invasive breast cancer without a prior diagnosis of DCIS. Further analyses of the women with DCIS will be conducted according to whether or not they were subsequently diagnosed with invasive breast cancer. Three groups ((i) DCIS patients (n=2000 women), (ii) age matched controls (n=8000 women), and (iii) primary invasive breast cancer patients (n=2000), ensuring at least 4 women as controls for each woman with DCIS or invasive breast cancer) will be included in this study with a total of approximately 12,000 women. After 12-18 months a further data request will be made in order to increase the sample size of the cohort further by including new incident DCIS cases.

Chronic musculoskeletal pain (CMP) is prevalent health issues that impact the quality of life of many individuals. CMP is defined as persistent or recurrent musculoskeletal pain deriving from musculoskeletal structures such as joint, muscles and bones and last for at least three months(Zhuang et al., 2022). It is commonly found among older adults, with a point prevalence of more than 50% among elderly (Yamada et al., 2022).

Constipation is also commonly occurs among patients with CMP. While it is known that constipation may lead to chronic back pain, evidence also suggests that constipation may affect gut microbiome, which in turn can affect pain perception (Arai et al., 2018; Smith et al., 2008).However, the current understanding on the association between CMP and constipation remains unclear.

In addition, constipation is frequently associated with colorectal cancer (CRC). In 2018, CRC was the fourth mostly commonly diagnosed cancer in the world, with approximately 2 million new cases (Rawla et al., 2019). However, for the severity of constipation or bowel movement frequency that could induce CRC, it remains unclear. By understanding the correlation between CMP, constipation and CRC, it could contribute to the development of comprehensive treatment strategies to improve the health-related quality of life (HRQoL) of these individuals.

Over the last ten years, polygenic risk scores have advanced from academic subject matter to clinically useful tools, capable of assisting diagnoses and setting screening standards. Yet, as medical professionals embrace the use of these tools in the US healthcare system, researchers have warned that these predictions are worse than expected for non-Europeans and cannot stratify risk within families as well as expected. Over the last several years, statistical geneticists have developed theory for and empirically analyzed the causes of these differences in prediction accuracy by ancestry.

Prediction accuracy tends to get worse as the genetic distance between training cohort and test cohort increases, and this can even occur when applying polygenic risk score models trained on a British cohort to an Italian test cohort. To help fix these problems, researchers have developed a number of tools that help construct risk prediction models that take ancestry into account.

Within the next 36 months, we aim to systematically assess these methods across a variety of traits to compare their utility risk prediction in non-European cohorts. We further intend to investigate whether these gaps differ by the trait or disease analyzed.

The proposed research fills an important gap in the current literature because there is no standardized baseline. Since differences in risk prediction by ancestry create a barrier to equitable use in the clinic, this research addresses an important need in public health.

Cardiovascular-kidney-metabolic syndrome (CKM) is a novel concept introduced by the American Heart Association (AHA) in 2023, which refers to a systemic disease resulting from the interactions between metabolic risk factors, chronic kidney disease, and the cardiovascular system. This syndrome can lead to multi-organ dysfunction and increase the risk of cardiovascular adverse events. However, the early diagnostic biomarkers for CKM have yet to be established, and the social and environmental factors involved in its progression remain unclear.

Objective
To address this gap, the present study aims to adopt an integrated approach that leverages extensive genetic, metabolomic, proteomic, and comprehensive epidemiological data. Through multidimensional analyses, this study seeks to identify novel risk factors and uncover potential biomarkers for CKM.

Scientific rationale
Probing circulating proteins provides unique opportunities to uncover novel biomarkers and improve our understanding of etiology of those diseases. Similarly, blood metabolome is considered as important readouts of aggregated information from genetic factors, gene expression to protein abundance, as well as external environmental factors. To accomplish this, we will combine multi-omics (i.e., proteomics and metabolomics) and epidemiological data to discover novel risk factors, biomarkers and provide definitive evidence for known CKM risk factors reported by traditional observational studies. Our endeavor will encompass the analysis of multi-omics and epidemiological information, enabling us to unravel new risk factors, identify potential biomarkers, and comprehend the intricate web of causal relationships underpinning CKM. This comprehensive exploration will encompass conditions such as diabetes, chronic kidney disease, atherosclerosis, vascular calcification, and others.

Neurological disorders such as Alzheimer’s disease, Parkinson’s disease, and stroke result from complex interactions among genetic, systemic, behavioral, and environmental factors. This project aims to explore how multi-organ imaging phenotypes, multi-omics data (genomics, proteomics, metabolomics, epigenomics), lifestyle behaviors, and environmental exposures contribute to the risk and development of neurological disorders.
We will address the following key questions:
How do brain and peripheral organ structural changes relate to neurological outcomes?
Which circulating biomarkers and molecular profiles are predictive of disease onset?
How do genetic and epigenetic factors interact with modifiable exposures such as air pollution, diet, and physical activity?
Can integrative models combining imaging, omics, and environmental data improve early disease prediction?
This project seeks to develop a systems-level understanding of neurological disorders, with potential applications in risk stratification, biomarker discovery, and targeted prevention strategies.

Traditional observational studies have reported positive associations among (a) type 2 diabetes, (b) chronic kidney function, and (c) coronary heart disease. However, whether the association is due to confounding or reverse causality is unclear. This proposal aims to investigate the causal association among these 3 health conditions with each other using Mendelian randomization instrumental variable analysis. The proposed research will help investigate how these diseases affect each other using Mendelian randomization design which is less susceptible to confounding than observational studies. The results will provide additional insights concerning the bi-directional association among 3 major health threats (cardiovascular disease, diabetes and renal disease) with corresponding implications for etiology and therapeutics. We compare risk of cardiovascular disease, type 2 diabetes and renal disease, according to genetically determined per unit of risk difference (per odds of) for these three diseases. 500,000 participants (full cohort) for the analysis involving cardiovascular disease, type 2 diabetes, renal disease and genetic data.

Global dementia cases are projected to exceed 80 million by 2040, with Alzheimer’s disease (AD) and cerebrovascular contributions representing major etiologies. Despite clinical co-occurrence, the genetic interplay and causal pathways between cerebrovascular disease (CeVD) and AD remain poorly characterized. Current evidence cannot distinguish causal relationships from shared risk factors, creating a critical knowledge gap for developing effective prevention strategies.
To address this gap, we will employ advanced genetic methodologies within the UK Biobank cohort. Our specific aims are to: (1) develop and validate polygenic risk scores for both diseases using a two-stage approach; (2) quantify shared genetic architecture via cross-trait PRS associations and pleiotropic locus identification; and (3) determine bidirectional causality using Mendelian randomization to establish evidence for causal pathways.
Genetic instruments enable causal inference in complex diseases. Polygenic risk scores integrate genetic susceptibility across variants to quantify disease risk, while Mendelian randomization utilizes these variants as natural experiments to establish causality and overcome confounding. Together, these methods provide powerful tools to elucidate disease etiology beyond conventional observational approaches. To solve this, our study will use genetic data as a powerful tool to uncover the true relationship. We will develop polygenic risk scores, which measure a person’s inherited risk for each disease. Using a method called Mendelian randomization, we will then treat these genetic scores as natural experiments to test if higher genetic risk for stroke actually causes a higher risk of Alzheimer’s, or if the reverse is true. The findings will provide crucial evidence on whether preventing or treating cerebrovascular disease can also reduce the risk of developing Alzheimer’s dementia, guiding future clinical strategies.

Considering the causes of Alzheimer’s disease, hereditary factors come to the fore for Early Onset Alzheimer’s Disease. However, the causes of Late-Onset Alzheimer’s Disease, which constitutes the majority of the patient population, have not been clarified yet. Age is the most prominent risk factor for Alzheimer’s Disease, but the effects of ageing on the disease are not fully understood. With ageing, mutations which are not acquired congenitally occur in blood cells. These mutations have been found to be less than 1% in people with age >40, about 10% in people with age >70, and about 15-20% in people with age >90 in investigations involving people without haematological malignancy. This condition has been termed clonal haematopoiesis of indeterminate potential. Individuals with these mutations are at risk of developing haematological malignancies such as Myelodysplastic Syndrome, Acute Myeloid Leukaemia, and Lymphoma. In addition, these mutations increase the risk of coronary heart disease, myocardial infarction and mortality. It has been discovered that these mutations, the prevalence of which increases with ageing, lead to some gene activations and inflammatory effects. We intend to use Whole Exome Sequencing data from the UK Biobank to investigate whether clonal haematopoiesis of indeterminate potential is a risk factor for Alzheimer’s disease in this study. The findings of this project will shed light on the aetiology and pathogenesis of Alzheimer’s disease.

Cancer, cardiovascular disease, diabetes (types 1 and 2), neurological disorders, and kidney disease present vast health and economic challenges. It is thus important to diagnose diseases early and tailor treatments precisely to the diverse needs of patients. In addition, no major disease exists in isolation, motivating a holistic understanding of the biology of each person.

We seek to use UK Biobank data to (i) identify molecular and physiological biomarkers predictive of disease predisposition, onset, and progression, and (ii) uncover cross-disease biosignatures that reveal shared biological connections.

Under Dr. Dimitris Bertsimas, our group has a proven track record in health AI. We demonstrated the strengths of holistic machine learning (ML) approaches, (i) developing a ML framework that integrates multimodal data (e.g. labs, notes, images) to reliably predict patient outcomes (Integrated Multimodal Artificial Intelligence Framework for Healthcare Applications, Soenksen et al., 2022), (ii) extending this to multitask across multiple clinical endpoints (M3H: Multimodal Multitask Machine Learning for Healthcare, Bertsimas & Ma, 2023), and (iii) writing an ML textbook on improving healthcare delivery (The Analytics Edge in Healthcare, Bertsimas, Orfanoudaki, & Wiberg, 2023).

We believe we can incorporate cellular-level data with outpatient visit data to create a more comprehensive and holistic biological picture of each patient. We aim to identify biosignatures for early, accurate, and minimally invasive disease detection, by integrating multiomics data and clinical phenotypes to identify clinical-molecular associations that can better model disease trajectories.

Through this unified ML approach, we aim to build interpretable data-driven models that inform early detection and personalized treatment strategies. Our expertise and experience uniquely position us to uncover both shared and distinct etiologies across these critical disease categories.

Human disease, is a complex trait that occurs once in a life, based on a biological basis, however its expression is variable, on intensity, onset and duration, because its personal genetic basis (gene-centred and genome-wide) and their environment. In this point one disease itself or mediated by secondary effects of it treatment, modifies the presentation of other conditions, that also based on their genetics (shared or not) and their environment, modify the occurrence of further diseases or their evolution. This is the rule for most of common diseases, chronic conditions that occurs once and that are treated chronically.
Our research plan is based on the time and order in which diseases are diagnosed in the population: First, we will apply different technical approaches to identify trajectories based on single diagnostics or aggregation analysis using cluster comorbidities, based on age-gender strata. Second, using genetic profiles (genome wide), we will analyse shared genetics among members of trajectories, o lead disease among clusters, to identify polygenic profiles that could identify temporary progression. PRS will be used as Instrumental variables. Adjusted variables will be used to account for impact of ethnic and educational level on disease diagnostics. Third, UKBB, will be used as a validation steep, using the same approach used in the Catalan cohort, to validate the results. The increased power of the UKBB database, will allow the exploration of additional observed comorbidities or temporary patterns that due to lack of power in the discovery cohort could not be explored in the first steep.
The identification of the genetic profile, as biomarkers or as etiological determinants, will help us gain a better understanding of the diseases and the health of the population, moving to an informed practice of the health-care management. This will impact the health-care system, with a preventive and more personalized management of diseases using agnostic genetic information. In addition, promising disease relationships would be derived for further investigated using more in depth approached to identify pleiotropic loci, as a step for a better drug repositioning strategy.

Non-communicable diseases (NCD), such as cancer, heart disease, type II diabetes mellitus, chronic respiratory conditions like asthma, and poor mental health, are some of the most common causes of death in the UK. The design of our cities play an important role in preventing these chronic diseases and have a significant consequent impact on the quality of life and life expectancy of their citizens. However, in a rapidly evolving urban age, urban designers, urban planners and public health practitioners still know surprisingly little about how best to design our cities in order to prevent NCD and their known risk factors.
Our aim is to generate evidence and tools to support the urban planning and health sectors to better understand how to design our cities to prevent NCD. Objectives include:
1. Use new methods in computer vision and artificial intelligence to explore the relation between urban design and NCD in cities across the UK.
2. Investigate how different designs within cities impact on health inequalities including NCD.
3. Combine data from different sources to investigate the mechanisms by which the design of our cities causes NCD.
4. Learn lessons about how different ways of designing our cities prevent NCD and their known risk factors.
5. Develop a toolkit for action for local citizens, urban designers and planners, public health practitioners and policy makers, to help inform future policies and lead to powerful, actionable changes in the city.
6. To build a legacy of transdisciplinary research capacity in public health science, urban design and computer science, with clear pathways to impact.
We will build the evidence base on the relationship between urban design and NCD. This will be comprised of two levels of analysis: the city level and individual level. This acknowledges that people’s health is affected by their immediate environment and by the way the entire city is organised. This will help us understand which features of the built environment are associated with city designs, NCD and their known risk factors, such as not being physically active, having poor diet, smoking, consuming alcohol and being exposed to air pollution. We will then test changes to the design of our cities to analyse how they will help prevent NCD and reduce known risk factors. The findings will help inform future policies and practices leading to actionable changes in the design of our cities to make them healthier places for people to live.

Aim: Diabetic kidney disease (DKD) develops in approximately 40% of patients who are diabetic and is the leading cause of end stage renal disease worldwide. Our previous study indicated that abdominal obesity is an important risk factor of DKD, consistent with several observational studies. However, the causal correlation between the abdominal obesity and DKD remains unclear.

Scientific rational: To identify the causal correlation between the abdominal obesity and DKD, we intend to use the approach of Mendelian randomization, which has traditionally been used to determine whether a candidate risk factor is causally related to a clinical outcome using the genotype as an instrument. In a Mendelian randomization study, genetic variants is used as to divide a population into genotypic subgroups, in an analogous way as how participants are divided into arms in a randomized controlled trial. By dividing participants into subgroups with different genetically exposures to the risk factor, we can evaluate the causal effect of the risk factor on disease risk. A recent large-scale Genome wide association study (GWAS) identified 48 SNPs associated with abdominal obesity, and combining these 48 SNPs into a weighted polygenic risk score, we can determine whether a genetic predisposition to abdominal obesity is associated with DKD.

Project duration: The duration of this project will be three years, from 2020 to 2023.

Public health impact: We expected to identify the causal association between abdominal obesity and the development of DKD. If further research validates our findings, two recommendations on the risk assessment and intervention of DKD might be considered by clinical practice: (1) Clinical screening of abdominal obesity by simple anthropometric measures to identify diabetic patients who might be more susceptible to diabetic nephropathy; (2) Interventions aiming to reduce abdominal adipose tissue may become an important adjunct to glycemic and blood pressure control for reducing the risk of DKD.

Research on environmental influences on cognition/ADRD has primarily focused on single exposures, suggesting air pollution, noise , limited greenspace , sunlight, and artificial nightlight exposure, are individually associated with heightened risks of ADRD, mental health problems, and sleep disturbances. In real world settings, however, people are often exposed to multiple adverse environmental factors simultaneously. These overlapping exposures may amplify health risks beyond the effects of individual factors. Understanding the aggregated impact of these exposures is essential for identifying high-risk combinations and evaluating the combined effects of multiple exposures on sleep and cognitive health in aging. This study will introduce an aggregated environmental exposure model to identify high-risk environmental combinations for sleep, mental health and AD/ADRD.

It is critically important to prescribe the right medications at the right time to get the best treatment results for patients. Our previous research has found ways to improve treatment strategies by understanding how different patients respond to the same treatments. This was done using information like patient demographics, medical history, and standard lab results. However, because cardiometabolic diseases (like hypertension, type 2 diabetes, and heart disease) have complex underlying causes, this clinical information alone may not be enough to fully understand a patient’s condition. To better personalize treatments, we plan to use advanced protein analysis (proteomics) to: 1) identify how different patients respond to treatments for cardiometabolic diseases, 2) understand the molecular reasons behind these different responses, and 3) determine the best use of each drug based on new proteomic biomarkers found in blood samples. By combining traditional clinical data with proteomic data, we hope to more accurately assess patient conditions and tailor treatments to improve outcomes.

The social environment, dietary habits, lifestyle, clinical comorbidities, and body metabolism play a key role in the global burden of metabolic cardiovascular disease. Therefore, identification and effective management of modifiable factors are essential for the prevention and control of CVD. Given the differences between populations, the importance of various risk factors varies across populations. Therefore, by utilising data from the UK Biobank, we plan to calculate group attributable risks for metabolic CVD risk factors in different age and genetic susceptibility groups. Subsequently, we will rank these risk factors according to group attributable risk to identify high-risk, modifiable factors. In addition, we will investigate how metabolic cardiovascular risk factors vary with individual age and genetic susceptibility. This study will build on existing evidence and support national and international guidelines for early intervention in the management of metabolic cardiovascular disease. Understanding the interactions between genetics and metabolic cardiovascular risk factors will contribute to more precise management of metabolic cardiovascular disease in the future.

Background:
Restless Legs Syndrome (RLS) is a neurological disorder characterized by discomfort in the legs while resting and an uncontrollable urge to move the legs to relieve the discomfort. RLS more typically occurs in older adults. RLS commonly follows a circadian rhythm, worsening in the evening and at night, delaying sleep onset. Thus, while older adults without RLS typically fall asleep and wake up earlier than younger adults, those with RLS often fall asleep later and have their best sleep in the early morning. This distinct pattern of sleep has been anecdotally noted by Dr. John Winkelman, MD, PhD.

We aim to compare sleep patterns of RLS and non-RLS patients, as measured by hour-by-hour actigraphic data. If a unique pattern of sleep is identified, distinguishing RLS patients from non-RLS patients, it can be used as a tool to aid in the diagnosis of Restless Legs Syndrome.

Research Question:
How do the sleep patterns of patients with Restless Legs Syndrome (RLS) differ from those without the condition, based on actigraphic measurements?

Objectives:
– Compare actigraphic data of sleep patterns in RLS patients and non-RLS controls.
– Investigate if RLS patients exhibit a distinct sleep pattern, particularly in the early hours of the night and morning.

Scientific Rationale:
Although the pattern of sleep disturbance in RLS has been noted anecdotally, there has been little systematic investigation into these circadian trends, especially using objective, quantifiable measures like actigraphy. This project aims to fill this gap by using actigraphy to compare the sleep patterns of RLS patients versus non-RLS controls, offering insight into the possible use of sleep disturbances as a diagnostic feature for RLS.

We aim to enable new forms of collaborative care for long-term conditions by integrating data from wearable sensors (accelerometery) with the prediction of health risks from electronic health records. This will let us create algorithms for adaptive, personalized care planning that takes account of individual predicted risks and real-time sensing. We will investigate serious mental illness (schizophrenia), chronic kidney (renal) disease, and controls. We will quantify episodes of activity data as different measures/risk factors, use this as an input to our predictive modeling, and contributes toward an understanding of whether activity monitoring can be used an indicator of remission/relapse. We are developing software tools which will help patients with long-term conditions, together with their carers and doctors, to better manage their health in daily life, respond more quickly to changes, and prevent fall back episodes. By identifying associations between unsupervised behavioral phenotype (mobility, rhythmic/routines, sedentary behavior, fitness/frailty issues, weight gain/loss) and disease progression stages, we will build new disease risk prediction models. We have a strong focus on translation, with a dedicated 3 year work package on health and economic benefit analysis, so we can advance diagnosis, prevention/early detection, and risk stratification for chronic kidney disease and schizophrenia. The analysis stage will identify indicators (i.e. patterns/phenotype/risk factors) in the accelerometer data that assign patients to different disease states.(?Unsupervised? machine learning methods will be used to identify patterns in the data itself, without any intervention from the user, or a human interpreter.) These accelerometer based indicators will be used as one input, together with data on hospitalizations, self-reported measures, blood samples, and similar, for developing risk prediction models of how likely each person is to relapse, or to be in remission, or to be in another identifiable state which might alter their care planning.

Chronic diseases such as diabetes, cardiovascular diseases, and cancers develop through multiple culmination of processes including lifestyle, dietary, environmental and genetic factors. The development of diseases occur on a continuum with the final stages resulting in detectable disease, and patients having symptoms. There are several stages prior to this, that we can detect early changes in the body, in which if intervention can be targeted, can be more effective at disease prevention and cure. This study will examine the early changes based on body MRI quantitative measurements and physical activity measurements. These changes can be subtle and will require comparison with population reference ranges which can now be calculated based on the data collected by the UK Biobank.

Our project aims to derive population normative range using the UK biobank data and the local Chinese cohorts with data collected in the same manner and to combine the various quantitative parameters for assessing individual wellness.

The idea is we may be able to create biomarkers for the cardiovascular, cancer-related and all-cause mortality prediction. The impact to the broader population will be that there will be a more accurate method of screening of subclinical disease detection based on these non-invasive data points opening a window for serial monitoring during the intervention to improve health and individual wellness. Accurate and reproducible biomarkers are desirable not only to improve how we can measure health but also in assisting large scale clinical trials. We anticipate the findings to significantly contribute to the advancement of preventative medicine. The project duration is expected to be 36 months.

Parkinson’s disease (PD), is one of the most debilitating neurodegenerative disorders characterised by motor symptoms including bradykinesia, tremor, and rigidity. While a complex interaction of genetic and molecular mechanisms has been postulated to play an important role for the vulnerability of brain networks to neurodegeneration, little is known about the metabolic and neuroinflammatory mechanisms that substantially modify disease courses. Recent work using single-cell RNA transcriptomes from cortical tissue of Parkinson’s patients has identified a strong association between disease progression subtypes and subsets of microglia, oligodendrocyte progenitors and astrocytes. Dysregulated mitochondrial electron transport, abnormal oxidative phosphorylation pathways and an inflammatory microglia response might be mainly involved. In this project we aim defining disease progression phenotypes (defined as age at disease onset, motor impairment at distinct disease durations and motor and cognitive worsening over disease course) to proteomic and metabolomic characteristics. We will then relate these to distinct charactersitics of neurons and glia subsets as derived from own sc-transcriptomes. Genetic and MRI variables will be considered as confound factors and further included in the multivariable analyses. We aim describing novel direct mechanistic links between neuroinflammation and neurodegeneration through mutual cellular subsets and gene proteomic and metabolomic expressions in PD for different disease subtypes can be targeted in future personalised therapeutic interventions.

Few studies have investigated potential risk factors for in situ breast cancer . Although the association between in situ breast cancer risk and few individual risk factors has been investigated, analyses on diet and lifestyle in relation to in situ breast cancer from large studies are missing.

The World Cancer Research Fund/American Institute for Cancer Research has recently published nine cancer prevention recommendations. Adherence to these recommendations has been associated with reduced cancer incidence and mortality, including reduced invasive breast cancer risk. However, to our knowledge, no study has investigated the association between adherence to the World Cancer Research Fund/American Institute for Cancer Research cancer prevention recommendations and in situ breast cancer risk.

In the analysis anticipated, each of the potentially modifiable World Cancer Research Fund/American Institute for Cancer Research cancer prevention recommendation will be scored, as in previous publications. The sum of the single recommendations will then constitute the score used to assess the risk of in situ breast cancer development. Since in situ breast cancer is mainly detected through mammographic screening, we will utilize the mammographic information in the UK Biobank (mammographic screening attendance, time since last mammogram) as covariates and stratifying variables in our analyses.

We aim to investigate whether adherence to the potentially modifiable World Cancer Research Fund/American Institute for Cancer Research cancer prevention recommendations is associated with in situ breast cancer development overall and by socioeconomic characteristics in the UK.

Coronary artery disease (CAD) is a complex traits with heritable, environmental, and lifestyle factors contributing to disease risk and disease course. Both, genetic as well as dietary factors are known to be associated with risk for new-onset (incident) CAD. However, data on the interplay of dietary patterns (a more comprehensive analytical approach to diet than single food analyses) and genetic variants (cumulatively expressed as genetic risk score) on the risk of incident CAD are lacking. We are applying for data from the UK Biobank to address this issue. Specifically, we are aiming to:
1. Derive a priori (Alternate Mediterranean Diet, Alternate Healthy Eating Index-2010, and the Dietary Approach to Stop Hypertension) and a posteriori (using PFA, PCA, RRR, and LCA) dietary patterns based on data from the UK Biobank;
2. Assess associations between these derived dietary patterns and incident CAD – as defined in (46) – within the UK biobank.
3. Assess whether the association of dietary patterns with incident CAD differs by genetic risk (based on a weighted genetic risk score).
4. Investigate influences of scoring alternatives on CAD risk.
5. Investigate changes of dietary patterns over time and their determinants and assess how these changes relate to CAD status alone and according CAD-GRS categories.
The present project will increase our understanding of the complex interplay between genetic factors and nutrition with respect to incident CAD. It might inform future personalized heart-nutrition strategies aimed at preventing CAD.
Full cohort required. Data request: FFQ and FFQ repeated measures, web-based 24-hour-recall questionnaires, clinical, anthropometric and CVD biomarkers data, family history of CAD, genetic data, CAD endpoints.

Virtually no epidemiologic studies have evaluated the association of objectively measured central obesity with cancer risk. The UK Biobank cohort provides an exceptional opportunity to evaluate this association. We propose: 1) to evaluate the associations of objectively measured body fats, weight gain, and BMI with total and site-specific cancer risk; 2) to determine associations of BMI and central obesity with breast and colorectal risks via Mendelian randomization analyses; and 3) to assess the association between GWAS-identified BMI-associated genetic variants and weight change over the lifecourse. The associations of both adiposity and adult weight gain with cancer risk are of increasing public health importance because the prevalence of overweight and obese adults has increased markedly in recent decades. The results of our proposed study will increase knowledge of the influence of obesity on cancer risk. Additionally, our Mendelian randomization study, along with the evaluation of BMI-associated variants with weight gain, will provide insight into the complex relationship of genetically-determined body weight and cancer risk. We will use statistical methods to evaluate the association of cancer risk with body fat, fat around the mid-section, and weight change from childhood to adulthood. We will evaluate these relationships using information collected on body composition during the participants? baseline visits, and by using information from genetic markers that are known to be related to body mass index. We plan to conduct a cohort analysis for the first aim, including all cohort members who provided the data needed for the analysis. For the second aim, a nested case-control study will include all cases of breast (2,320) and colorectal (1,256) cancer and 4 matched controls per case whose genotyping data are available (approximately 17,880 total). The third aim will utilize all participants with sufficient genotyping data, a baseline measure of BMI, and at least one of the following measurements: a value for comparative body size at age 10 or a repeat assessment value for BMI.

Living with multiple long-term health conditions is common and important for both patients and for the NHS (National Health Service). The way we deliver health services in the NHS is not ideally suited to care for people with multiple long-term conditions. To prevent people from acquiring multiple long-term conditions and to provide the best care to people with multiple long-term conditions we urgently need to better understand how to prevent and treat different groups of long-term health conditions that tend to occur together, known as clusters of long-term conditions.

As part of our 4 year collaborative programme of research, funded by the Medical Research Council and National Institute for Health Research we plan to use data from UK Biobank to address this need by:
1) establishing how the area where a person lives and other socio-demographic factors are linked with different clusters of long-term conditions. We will also look at whether lifestyle factors such as diet and exercise that may have a role in explaining any links we find
2) gaining new understanding of what processes lead to the development of different clusters of long-term conditions, using the genetic data available in UK Biobank.

This work will lead to the development of robust evidence that will provide the basis for novel public health, policy and biopsychosocial interventions to prevent, treat and mitigate the consequences of different clusters of long-term conditions.

It is hypothesised that carcinogenic viruses could be transmitted through breast milk and cause breast cancer. There might be other differences between babies fed with breast milk and with bottle milk that contribute to cancer in adulthood, such growth hormones or gut microorganisms. Whether there is difference in adult cancer risk between babies who are breast-fed and were bottle-fed remains under-investigated. We propose to compare risk of adult cancer risk and related diseases between those who were breast-fed and those who were not breast-fed when they were babies among UK Biobank participants. One purpose of the UK Biobank is to ?support a diverse range of research intended to improve the prevention, diagnosis and treatment of illness and the promotion of health throughout society.? The proposed research aims to understand whether breast feeding is associated with cancer risk of specific sites and risk of related diseases. The results may provide hints to preventable causes of cancer. At the recruitment of the UK Biobank, participants reported a wide range of information, including whether they were breast-fed when they were babies. They also gave consent so that their ongoing history of hospital admissions and cancer diagnosis after recruitment could be obtained for research. Among participants who did not have cancer at the recruitment, we will compare risk of developing newly diagnosed cancer and related diseases between participants who reported they were breast-fed and not breast-fed when they were babies in the UK Biobank. The full cohort is requested for this study.

Genetics, socioeconomic status, education, neurobiology and medical history all interact in complex ways to confer sensitivity or resilience to central nervous system insult. For example, even mild TBI, in individuals with high CNS frailty, may lead to increased risk for stroke, neurodegenerative disease, anxiety or depression later in life. The traumatic brain injury (TBI) research group at the University of Virginia, in collaboration with Cohen Veterans Bioscience, proposes the use and analysis of the neuroimaging, genetic, cognitive testing, medical history, and other demographical data available within the UK Biobank repository to study the linkage between neurological and cardiovascular health to better understand the diverse outcomes (i.e. the sub-populations) represented in military veterans and other aging populations. These data and resulting analyses will inform potential precision medicine approaches, guide future research avenues, and further refine the development of open-source tools to conduct such research.

The goal of this research project is to develop and refine machine learning models to better identify individuals at risk of developing diabetes. Specifically, the project aims to:
1. Detect individuals with prediabetes.
2. Stratify these individuals into low, intermediate, and high-risk categories for progressing to diabetes.
Scientific Rationale:
Prediabetes is a condition where blood sugar levels are higher than normal but not yet high enough to be classified as diabetes. Early detection of prediabetes is crucial because it allows for timely interventions that can prevent or delay the onset of diabetes. Diabetes is a serious condition that can lead to severe health complications such as heart disease, kidney failure, and blindness.Traditional methods of diagnosing and assessing the risk of diabetes have limitations, such as relying on a few specific biomarkers and not accounting for the complex interactions between different risk factors. Machine learning, a type of artificial intelligence, can analyze large amounts of data and identify patterns that might not be apparent to human researchers. By leveraging electronic health records and comprehensive datasets, we can develop more accurate and personalized models for predicting who is at risk of progressing from prediabetes to diabetes.
Project Duration:
The project is expected to last three years, including the phases of data collection, model development, validation, and implementation.
Public Health Impact:
This research has the potential to significantly improve public health by:
1. Enhancing Early Detection:** More accurately identifying individuals with prediabetes, allowing for earlier and more effective intervention strategies.
2. Personalizing Care:** Stratifying individuals based on their risk levels helps tailor prevention and treatment plans, making them more effective.
3. Reducing Healthcare Costs:** By preventing the progression to diabetes, the project can help reduce the long-term healthcare costs associated with treating diabetes and its complications.
4. **Improving Health Outcomes:** Ultimately, the research aims to improve the quality of life for individuals at risk of diabetes by preventing the onset of the disease and its associated health issues.

Cardiovascular diseases, such as ischemic heart disease, hypertension, stroke, heart failure, peripheral artery disease, etc., are among the leading causes of mortality among adults. Identifying risk factors for these diseases is crucial for preventing or delaying complications and premature death.
Aims:
1. To comprehensively assess proteomic, metabolomic, genomic, and other omic factors, as well as imaging, environmental, lifestyle, psychosocial factors, etc., within the UK Biobank and their causal associations with cardiovascular diseases and comorbidities.
2. To evaluate the associations between trajectory changes in omic and non-omic risk factors, with the risk and mortality of cardiovascular diseases and multi-morbidity.
3. To develop risk prediction tools (e.g., models, scores, algorithms, etc.) that incorporate omic and non-omic factors for cardiovascular diseases and multi-morbidity.
4. To utilize advanced statistical methods (e.g., machine learning) to improve the accuracy and efficiency of identification and risk prediction of cardiovascular diseases and their comorbidities.
Scientific Rationale: Although numerous factors affecting the risk of cardiovascular diseases have been identified, many more remain undiscovered. Non-omic factors, including dietary habits, exercise behavior, sleep patterns, nutritional status, etc., are well known to influence these diseases. Additionally, the interactions between proteomic, metabolomic, genomic, and non-omic factors must be considered. Moreover, there is a lack of sufficient and robust evidence on estimating the risk and mortality of cardiovascular diseases by incorporating the trajectory changes of these factors in a population-based prospective cohort. The integration of advanced statistical methods and phenotyping techniques can further enhance the identification and prediction of cardiovascular diseases by uncovering complex patterns and relationships in the data. A better understanding of the factors affecting the likelihood of developing cardiovascular diseases is essential for creating novel and more effective approaches to disease prevention, diagnosis, and treatment. This project is expected to last for 36 months.
In addition, several other diseases (including metabolic, respiratory, digestive, neurological, renal diseases, and cancers) are known to have significant links and interactions with cardiovascular health. By comprehensively analyzing data on these related diseases, we can better inform the development of improved prevention strategies, targeted interventions, and personalized treatments for individuals at risk of multi-comorbidity.
Public Health Impact: Our study will enhance the understanding of the etiology of cardiovascular diseases and provide scientific evidence for early intervention and improved disease management. This will ultimately contribute to better public health outcomes by informing strategies to reduce the burden of cardiovascular diseases and their comorbidities.

Our project combines computer science and public health expertise to explore and enhance health risk prediction models using data from the UK Biobank. We aim to understand better the complex relationships between lifestyle choices, environmental factors, and other variables and how they impact health outcomes such as mortality, diseases, hospital visits, and cardiovascular events. Traditional health risk assessments often look at isolated factors without fully accounting for how these factors interact. Our approach seeks to fill this gap by analysing the interconnections among various health risk factors, thus improving our ability to estimate their combined effects on individual and population health.
Concurrently, we are investigating the effectiveness of a gamified digital health intervention through a separate, ongoing Randomised Controlled Trial (RCT). This trial examines how gamification, implemented via the YuLife mobile health app, can motivate users to adopt healthier behaviours. The insights gained from the UK Biobank will complement the findings from the trial, enabling us to assess how lifestyle modifications influenced by gamification impact significant health outcomes. Our research aims to identify the risk factors most responsive to gamified digital interventions and predict the long-term impact of these interventions on health. This is particularly relevant in today’s world, where lifestyle-related chronic diseases are becoming more common. By combining the comprehensive biomedical data of the UK Biobank with the practical interventions tested in the YuLife app, we hope to establish a solid scientific basis for developing engaging, effective, and personalised public health strategies. The ultimate goal of our two-year project is to develop actionable, evidence-based recommendations that can be used by health practitioners, policymakers, and individuals to reduce health risks and enhance quality of life. These recommendations will focus on identifying and implementing effective and accessible digital health intervention strategies.

In this proposal we focus on estimating the risk of developing the disease (e.g. breast cancer) over time given risk factors. Survival analysis is a branch of statistics for analysing the time until a pre-specified event happen, such as death or a disease. Hence, it provides a useful tool for such prediction analysis while taking into account competing risks (i.e., a person who died before having the disease cannot have the disease anymore), and left truncation (i.e., only individuals who survive at least by age 40 are included in the UK Biobank dataset). Heritability summarise the proportion of the variance of the trait under study (e.g. disease status) that is due to genetic factors. Predictive performance of a model strongly depends on the extent of heritability of the trait. For any given sample size, more accurate prediction is possible for more heritable traits, such as Crohn disease and type 1 diabetes, than for less heritable traits such as prostate cancer. Improving heritability estimation for various traits could provide insights on several missing heritability questions and lead to practical solutions for improving risk-prediction models.
Finally, we will study complex and high dimensional traits. Current methods, which follow a two-step approaches: first, extract the relevant features from the high-dimensional traits and then perform genome-wide association analysis on these features. In contrast to these methods, we will apply a joint modelling approach where we extract the most heritable features by correlating them with the genotype data, which will enable us to discover new heritable traits, and may lead to changes in definition of relevant traits from high-throughput digital trait data.

Rationale:
The sequencing of the human genome offered a lot of promise for understanding the genetic basis behind many traits that are consequential for health and wellness. By understanding the genome, we might be able to understand why certain individuals suffer from specific illnesses, which could offer insight into how to manage and treat certain diseases. While decades of research have revealed a very strong genetic basis for many traits, questions remain about how certain aspects of our environment or experience may play a role in how we study genomes. More advanced statistical techniques may allow us to better measure how the environment may influence disease risk.

Aim:
We aim to study how many under-appreciated features of the environment-including those associated with social stratification (e.g., socioeconomic status)—can influence how we study and interpret the relationship between genetics and observable traits. We will develop and apply new statistical methods across a number of traits and environmental factors, learning how certain environmental factors may interact with and affect the genetic architecture of behavioral and psychological traits. The UK Biobank contains hundreds of thousands of individuals from diverse demographic backgrounds, which makes it an ideal set to examine these questions. It will allow us to examine many traits of interest and examine different environmental factors. Gene-by-environment interactions have been well studied in other complex traits (e.g., height). We plan to evaluate the power and robustness of our developed methods on these phenotypes, and then focus on behavioral and psychological traits, which can be influenced by experiences, and contribute to mental health conditions.

Duration:
Because the foundation for the statistical methods that we will apply for this work are published, we anticipate a brief study time. We anticipate 12 months from the start of the study, to study completion, where analyses will be complete, with the team starting the process of sharing results with the scientific community via publication shortly thereafter.

Public health impact:
While the science of genomics has already provided insight into how genes can influence disease risk, there remain many other factors that can influence disease risk. For example, the lived experience and exposure to certain environments can lead to certain mental health ailments. Our study hopes to propose better methods for separating the effects of genes from those of the environment. Further, we hope to identify some specific factors that might make some individuals more susceptible to mental illness.

While physical activity and good sleep is beneficial for human health both physically and mentally, most evidence of such claim is based on self-assessed measures of activity. In the UK Biobank study, data from activity monitors are collected 24/7 for 7 days from over 100K participants producing over 15 trillion movement readings. Coupled with extensive demographic and health information, this dataset will give us the first glimpse of baseline biometrics from healthy individuals as well as individuals experiencing various medical conditions.

Conventional algorithms to detect sleep from actigraphy data require static sleep diaries or actigraphy marker button to label the boundaries of time in bed. These manual practices add to the uncertainty and participant burden, which we aim to reduce by automating detection of sleep boundaries during both night sleep and day time napping from the data directly. Here, we aim to explore probabilistic models for sleep detection to improve upon the conventional algorithms that only distinguish binary sleep versus waking time. Finally, we aim to translate these innovations into open source software feasible to be re-used by non-domain experts.

Collaborating with Dr. Vincent van Hees, the Netherlands eScience Center in Amsterdam, the Netherland, we will begin a two year study to develop the algorithm and apply it in UK Bioback dataset to derive sleep and activity metrics in medical conditions to serve as benchmark for Lilly clinical studies planning to incorporate activity monitor as quality of life assessment tool.

Leveraging public datasets collected in the last few years by large observational studies such as the Whitehall II Study (Sabia et al., 2014), UK Biobank (Doherty et al., 2017), and the Newcastle 85+ study (Anderson et al., 2014), the project would advance the activity monitor data open source software development and accelerate public and private efforts to study impact of sleep health and physical activity on healthy living, aging, disease, as well as transition from wellness to illness, and possibly illness back to wellness with the aid of lifestyle adjustment or therapeutics

Personalized risk assessment is recommended in cardiovascular medicine guidelines. Scores are often used to individually adapt prevention, diagnosis and treatment. However, the implementation of this risk assessment in everyday clinical practice in Germany is inadequate due to major obstacles: the lack of relevant structured information; inadequately standardized and incomplete storage in electronic health records; Lack of interfaces and data linkage to enable rapid assessment and ultimately visualization. The same applies to high-resolution biosignal analysis, which represents an important untapped resource for risk assessment. We are researching novel concepts to combine score and biosignal-based risk analyzes and define their predictive performance in real clinical settings.

Chronic myelomonocytic leukemia (CMML) is a rare and aggressive blood cancer that affects the production of certain white blood cells called monocytes. Patients with CMML often have a poor prognosis, with limited treatment options and a high risk of progression to more severe forms of leukemia. Despite recent advances in our understanding of the molecular changes associated with CMML, the underlying genetic causes of the disease remain largely unknown. This lack of knowledge has hindered the development of effective prevention, diagnosis, and treatment strategies for CMML.

Our research project aims to address this challenge by leveraging the resources of the UK Biobank, which is a large-scale biomedical database containing genetic and health information from 500,000 individuals. By comparing the data of individuals with CMML to those of healthy individuals in the UK Biobank, we hope to identify specific variations that may contribute to the development and progression of CMML.

To achieve this goal, we will employ state-of-the-art computational methods to analyze the vast amounts of genetic and health data available in the UK Biobank. This will involve using machine learning algorithms to identify patterns and associations between genetic variations and CMML risk, as well as developing predictive models that can assess an individual’s likelihood of developing the disease based on their genetic profile, biomarkers, proteins, and other health factors.

The expected duration of our research project is three years. During this time, we will work closely with clinical collaborators and patient advocacy groups to ensure that our findings are clinically relevant and can be translated into meaningful benefits for patients with CMML.

Complex traits and diseases typically arise from the interplay of various elements, such as genetic, environmental, and lifestyle factors. Effectively addressing these diseases often demands a comprehensive approach that takes into account genetic predisposition, lifestyle modifications, and, when necessary, medical interventions. The rapid advancement of today’s technology is instrumental in studying complex traits using high-dimensional datasets, offering innovative frameworks to unravel the intricate connections between biomolecules, biological phenomena, and the environment. These approaches have been applied to numerous clinical studies, advancing disease diagnosis and enhancing treatment options for patients.
The UK Biobank stands as a treasure trove of data, encompassing genetic and lifestyle information about a substantial population. By integrating and analyzing extensive data from the UK biobank and our own collections, we aim to gain deep insight into the genetic basis of complex traits and establish related prevention and control measures.
We seek to 1) investigate the underlying mechanisms of complex traits and the intricate interactions among multiple factors. 2) explore the relationship between complex traits and other disorders. 3) intergrate imaging phenotypes with multi-omic biological data to develop sophisticated deep learning systems and models. These will aid in the accurate identification of clinical subtypes or drug resistance, prediction of effective treatment strategy, and identification of predictive biomarkers.
Project duration: Estimated duration of this project is expected to be 3 years.
Output: Our intended output over the next 2-3 years is to identify several biomarkers or reveal mechanisms that could lead to 1-2 publications in prestigious journals such as Nature Genetics, Nature Medicine, and Nature Communications.

Research questions
This research aims to advance the early detection and mechanistic understanding of neurological critical diseases, as well as related cardio-cerebrovascular and systemic conditions, by addressing the following aspects:
1. Identifying key clinical, biological, electrophysiological, genetic, and lifestyle factors associated with the onset, progression, and outcomes of neurological critical diseases
2. Developing deep learning models to predict disease risk, early manifestations, and critical events using large-scale, multi-dimensional cohort data.
3.Utilizing large language models (LLMs) to generate automated, personalized diagnostic and prognostic reports, making predictive insights accessible and clinically actionable.
Research objectives
1. Develop predictive models using deep learning and diverse cohort data (electrophysiological signals, genetics, biomarkers, imaging, clinical records) to identify early indicators of neurological critical and cardiovascular diseases.
2. Improve risk stratification by integrating lifestyle, medical history, and multi-modal biomarkers with primary data sources.
3. Automate clinical reporting with LLMs to generate personalized diagnostic and prognostic reports, supporting efficient and informed care.
scientific rationale for the research
Neurological, cardiovascular, cerebrovascular, and sleep-related diseases are major causes of global morbidity and mortality, yet early detection remains limited. The UK Biobank offers large-scale, high-quality data from over 500,000 participants, including electrophysiological signals, genetics, imaging, biomarkers, and clinical records. This project will apply deep learning to identify early indicators, improve risk stratification, and generate mechanistic insights. Large language models will translate predictive outputs into automated reports, supporting precision medicine and reducing the public health burden of these diseases.

We will investigate how variation in the genome influences diseases in the UK Biobank dataset. To understand how diseases work and what genes and proteins affect a human disease or trait, researchers search for relations between gene mutations and diseases. While currently, researchers look for associations of genomic areas or genes with traits in GWAS and TWAS methods, the results are not always accurate. One reason for the inaccuracies might be that a predicted change in mRNA abundance does not necessarily cause a change in protein abundance. So, taking this into account, to improve and extend these approaches, we plan to integrate additional data into the studies, mainly proteomics data. That will elevate the current results by providing another set of associations: between proteins and traits. The method will allow us to study the UK Biobank data from a different perspective and compare the protein-level associations with the ones from GWAS studies to prioritize and complement meaningful relationships between genetics, proteomics, and disease. We will carry out the research project on the entire set of UK Biobank samples. With this, we hope to provide important insights into disease and trait mechanics

In recent years, genomic studies have been able to identify genetic variants associated with important complex traits which can inform precision medicine. However, these studies have mainly included people of European ancestry, and individuals of other ancestry are still greatly underrepresented in genetic research. Diversifying genomic studies is critical to ensure that the benefits of these findings are shared beyond European populations.

We aim to build and test tools to better study the genetics of complex traits in diverse populations. This addresses two main barriers that currently prevent broader inclusion in genomics: the lack of analytical methods that are tailored to diverse ancestries, and the lack of analyses conducted that include diverse populations. Our work will improve the clinical utility of large-scale data-collection efforts and begin to address the concerning health disparities that exist across populations.

Our project aims to advance scientific knowledge by studying the association between patient specific characteristics in disease states and underlying associations with novel genetic mutations. The disease states we plan to study include aneurysms, arrhythmias, cardiomyopathies, and clotting disorders. The key aim of these projects will be to verify novel discoveries across a much larger sample, as contained in the UK Biobank. In the medical and scientific community, this type of research is known as phenotype to genotype correlation and has been well established as a method to advance knowledge of disease states.

The public benefit of our research may eventually lead to enhanced diagnosis, and more appropriate treatments of the conditions which we will study.

The overall aim is to examine the independent and combined associations between obesity, MVPA, sedentary time and CRF with all-cause, cardio-vascular and cancer mortality.
1. To examine the independent and combined associations of MVPA, obesity, sedentary time and CRF associated with mortality?
2. To examine whether physical activity modify the association between obesity (adiposity) and mortality?
3. To examine whether sedentary time mediate the association between obesity and mortality? Obesity, moderate and vigorous intensity physical activity (MVPA), sedentary time and cardio-respiratory fitness (CRF) are all associated with a number of various health outcomes and mortality. However, the combined associations between these lifestyle variables and mortality is unknown. Further, it is unknown whether objectively measured PA may modify the association between adiposity (BMI and waist circumference) on mortality and whether objectively measured sedentary time may mediate this association. We will quantify how obesity, physical activity sedentary time and fitness are associated with mortality and the relative importance of physical inactivity, obesity, high sedentary time and low fitness on mortality. This will enable important information for intervention strategies and policy makers. Full cohort n=500,000 for most research questions, n~100,000 for the analyses involving cardio-respiratory fitness and objectively measured physical activity.

Understanding how drug response to a given therapeutic treatment differs in patients is one of the most important and long-standing challenges in personalised medicine. For example, 30 – 50% of patients with major depression do not respond to their first antidepressant drug prescription, and adherence remains often very poor. While a great deal of effort has been placed on linking variation in drug transport and metabolism with treatment responses, very little is known about how genetic variability in receptor drug targets affects treatment response. Elucidating the spectrum and impact of how a genetic receptor variant influences ligand binding and/or signalling and, ultimately, therapeutic effect is vital and would serve as an important step for understanding variability in drug response.

The G protein-coupled receptor (GPCR) super-family spanning ~800 members (~4% of the human genome) allows environmental and physiological messages – communicated by distinct signalling molecules – to be relayed into adequate cellular responses. Apart from involvement in almost all physiological processes, GPCRs also mediate the effect of ~34% of drugs (Hauser et al., 2017 Nat Rev Drug Discov). In this project we focus on drugs used to treat schizophrenia, affective disorders and attention deficit hyperactivity disorder (ADHD), which prevalently target GPCRs directly or indirectly. Despite the significance of GPCRs and genetic variations for CNS drug therapies, personalised medicine is yet beyond reach, because we are missing integrated large-scale data showing which genetic missense variants affect the intended therapeutic effects and how.

Here, we will employ a unique ‘pharmacogenomics’ research strategy, which integrates receptor genetic variants with 3D structures, pharmacological experiments, patient registries and literature data. We first map and select genetic variants in GPCR drug targets affecting CNS drug response through computational models. We then determine how psychiatric drug responses are affected by genetic receptor target variations in cell-based assays. Those experiments are labor-intensive, as we need to capture all possible intracellular responses. We hence envisage that the project would approximately take 3 years. We expect to correlate similar molecular receptor phenotypes induced by genetic variation with clusters of prescription changes and reported adverse reactions. To test these hypotheses, we will integrate information on hospitalisation records and prescribed medicines within the UKBioBank cohort.

A successful project would provide evidence for genetic variants linked to therapies and would hence allow us to:

i) better understand treatment adverse reactions in psychiatric disorders
ii) personalise medicine prescriptions based on GPCR genotypes, and
iii) prioritise drugs for pharmacovigilance investigations

Many people have more than one long-term disease (known as multimorbidity). For instance, there are many people who have both heart disease and diabetes. There exist computer tools that help doctors to predict the risk that a patient will develop diseases in the future (for instance, heart disease). However, these tools always focus on single diseases. This is not helpful when we try to prevent and treat multimorbidity. In this project we will develop tools that predict the risks that a patient will develop multiple diseases (and what those diseases are).

Working out how best to treat patients with multimorbidity is not straightforward. Typically, each disease comes with its own course of treatment, but it is not clear how to combine these. Therefore, we will also develop tools that predict what would happen if a patient were given a particular treatment. Such a treatment could involve changing their lifestyle (e.g. stop smoking), taking a particular drug, or a combination of these things.

People with multimorbidity often end up taking many drugs (known as polypharmacy). This can be a problem because some drugs do not work when they are taken with other drugs. We will therefore also extend the computer tools such that they can predict what would happen if a patient, who is living with multimorbidity and is taking multiple drugs, would change their lifestyle or change the drugs that they are taking.

We will focus on patients with diabetes, heart disease, and related diseases.

Psychiatric disorders are a significant public health problem, accounting for a significant portion of disabilities worldwide. Studies have shown that many of these disorders have a strong genetic component, with heritability estimates for disorders like schizophrenia and autism being as high as 80%. Despite this, the genetics of these conditions are still not fully understood, with the majority of genetic risk factors yet to be identified.

This project aims to better understand how genetics affects psychiatric and neurodevelopmental disorders, with a particular focus on autism, ADHD, bipolar disorder, major depressive disorder, schizophrenia, and substance use disorder. To achieve this, we will apply state-of-the-art analytical approaches to data from the UK Biobank and Danish population-based case-cohort sample for genome-wide association studies (GWAS), whole-exome sequencing (WES) and whole-genome sequencing (WGS) studies. The goal is to identify novel genetic risks and advance our knowledge of the genetic architecture of these disorders, with a focus on the role of common and rare genetic variants, genetic differences in sex, and genetic differences across subgroups.

The long-term goal of this project is to improve the diagnosis and treatment of psychiatric disorders by utilizing the knowledge gained from these studies. To achieve this, we aim to construct a prediction model that incorporates various types of data such as genetic data, electronic health records (EHR), imaging data, and environmental data. By doing so, we hope to improve the accuracy of prediction and facilitate the early detection of psychiatric disorders, particularly when they exhibit similar signs and symptoms.

Coronary heart disease (CHD) and heart attack is caused by a combination of inherited and lifestyle/environmental factors. 17,000 new cases of CHD are expected to develop in the UK Biobank participants by 2017. The aims of our research are use the genetic data being generated in UK Biobank to identify genes that affect risk of CHD, investigate whether specific genes and certain lifestyle factors such as smoking interact to increase risk, determine whether adding genetic information can improve prediction of CHD, and identify causal mechanisms for CHD that can be targeted to develop new treatments. UK Biobank was established to improve understanding of the causes of common diseases and particularly the interaction between genes and environment and life-style factors. CHD is the commonest cause of death and disability world-wide. It is the archetypal disease caused by an interaction between inherited and environmental/lifestyle factors. Our research will improve understanding of the genetic causes of CHD, how these interact with environmental/lifestyle factors and how the findings can be applied to improving prediction, prevention and treatment of CHD. We will divide the UK Biobank participants into those with CHD (cases) and those without (controls) and compare their genetic information generated using the UK Biobank array to identify genetic variants that are associated with development of CHD. We will use lifestyle/environmental information collected on participants to see if there is an interaction between such factors (e.g smoking) and genes in increasing CHD risk. We will investigate whether adding genetic information to current methods can improve prediction of development of CHD. We will use genetic approaches to define the best targets to develop drugs against to tackle CAD We plan to use the full cohort for our studies.

Comorbidity is commonly defined as at least two conditions coexisting in the same individual, which has posed a great challenge to the healthcare system worldwide and created substantial effect on individuals, their families, and society. The mechanisms of comorbidities are closely intertwined, with multidirectional relationships, shared risk factors, and common therapeutic targets. These results demonstrate the need for a comprehensive assessment of comorbidities to establish comprehensive disease models to promote the development of precision medicine.
However, the shared genetic architecture and biological mechanisms underlying comorbidities development are still unclear. Most previous studies have focused on observational studies, which were affected by confounding factors and cannot reflect the direct causal relationship between exposure and outcomes. In addition, most of these researches mainly focus on three elements: clinical characteristics, auxiliary examination results, and demographic characteristics, ignoring the impact of genetic factors on the disease. Therefore, it is necessary to further explore shared genetic architecture and deeper mechanisms of the comorbidities, which may contribute to the understanding, prediction, prevention, and treatment of the diseases.
This proposal aims to explore how comorbidities affects each other using Mendelian randomization design which is less susceptible to confounding than observational studies. We also aim to find some potential drug targets through the investigation of genetic architecture and their biological mechanisms and to apply them to the prediction, prevention, treatment, and management of diseases.
The project will last for 36 months. This proposal aims to investigate the causal association among comorbidities with each other using Mendelian randomization instrumental variable analysis. The findings will provide additional insights concerning the bi-directional association among comorbidities, which may improve our understanding of the pathophysiological mechanisms and help identify novel biological pathways and therapeutic targets for improving the prediction, prevention, and treatment of the diseases. By improving the diagnosis and treatment of co-morbidities, people can be encouraged to seek timely medical help, and the financial burden on the healthcare system and society can be released.

Why we are doing this research: Adverse childhood experiences (ACEs) are stressful events that happen to people before the age of 18. They include things like abuse, neglect and parental separation. Previous research has shown that experiencing more ACEs as a child leads to poorer health as an adult. This might be because specific conditions (like chronic pain) are more severe. Or it might be because people who have experienced ACEs develop multiple long-term health conditions that are hard to deal with at the same time. We want to test the theory that people who have had ACEs may respond differently to pain medications (such as morphine), compared with those who have not had an ACE.

Our approach: People with lived experience of ACEs, chronic pain and multiple long-term health conditions helped us to decide on our research questions. This helps us ensure our work can explore topics that are meaningful to the relevant people.

Some example questions:
Compared with people who have not experienced ACEs, are people who have experienced ACEs more likely to…
1) … be prescribed pain medications?
2) … be prescribed higher doses of pain medications?
3) … experience harms relating to pain medications (such as addiction, overdose or death)?
4) … find pain medications unhelpful?
5) … have side-effects from pain medications?
We will see if the answers to these questions are different for people with multiple long-term health conditions. We will use the UK Biobank database to answer these questions. We think that the large size of this database (over half a million people) will make our results relevant to patients and clinicians.

Project duration: 36 months.

Public health impact: Knowing how ACEs affect our responses to pain medications is important. It may help some people understand how their childhood affects their current health. For many people, knowing “why” can help them come to terms with their condition. This research will highlight the importance of trauma-informed care, and may guide patients and healthcare professionals when discussing treatment plans. It may be that for some people pain medications don’t help, and that other options (such as psychological therapy) might be better.

Prolonged sitting can have serious health consequences and a high level of fitness does not protect against the negative effects of sitting for too long. Sedentary job poses therefore a risk to all individuals regardless of their physical activity outside work. We aim to identify occupations at high risk of prolonged sitting and advance our understanding of the health risks associated with sitting for extended periods of time. We will evaluate the impact of lengthy sitting on the risk of developing cardiovascular disease and/or type 2 diabetes. A score evaluating risk of prolonged sitting for a given occupation will be derived for future research. Self-reported and objective accelerometer-measured sedentary behaviour data will be used to examine health risks associated with prolonged sitting. Sitting in the workplace is of primary interest but different sedentary behaviours and their interactions with physical activity will be taken into account, with adjustment for socio-demographic, genetic, and behavioural factors. We expect this project to take 24 months.

Cardiovascular drugs, such as aspirin and statins, are effective in preventing and treating cardiovascular disease (e.g. heart attacks and stroke). However, their potential effect on other chronic diseases (e.g. cancer, osteoporosis, respiratory diseases, dementia, and Alzheimer’s disease) remains unclear. The aim of this project is to investigate the effect of cardiovascular drugs on complex chronic diseases. We will use data from the full UK Biobank cohort. And innovative statistical methods, such as matching method and negative control method, will be used to increase the credibility of results. This study is expected to last for 36 months. The findings will provide evidence-based guidance for healthy cardiovascular drug use.

Aims:
We aim to use the data available from the UK Biobank to understand how genetic, metabolic, and clinical factors affect the intersection between adverse pregnancy outcomes and cardiovascular risk in women.
Background:
The genome is made up of 3 billion base pairs – A, C, T & G – and variation in the genetic code has been identified to be associated with risk for many diseases. Certain cardiometabolic traits and medication target loci have been associated with risk for hypertensive disorders of pregnancy. Expanding off of this, it will be valuable to look across many other APO (phenotypes) to determine other important associations that may exist between cardiometabolic traits and medication target loci with all APOs, not just hypertensive disorders of pregnancy.
Duration:
The anticipated project should take approximately three years after obtaining access to the data.
Public health impact:
Despite advances in genetic and other molecular techniques, determination of which patients are at most risk of for a broad array of APOs has yet to be explored. This project may potentially identify novel associations of diseases associated with the genetic, metabolomic, and clinical areas we plan to focus on. Ultimately, we hope this work will lead to more personalized approaches to pregnancy and postpartum care and improved maternal morbidity outcomes. Understanding the interaction between APOs and CV risk will help us to identify women at risk, better estimate their long-term CV risk, and improve our efforts to prevent and manage heart disease aggressively.

Cardiometabolic diseases are the leading cause of death and morbidity worldwide. In number of deaths, they are followed by their complications, such as kidney disease, heart attack and stroke as well as neurological conditions, such as Alzheimer’s disease. Together, they are responsible for the most deaths and the highest costs in the world in terms of treatments and hospitalisation. These conditions are caused by both genetics (inherited DNA) and environment (diet, activity, sleep).

In our research we aim to investigate (i) critical genes, regulatory regions and their interactions that influence risk of the disease; (ii) the complex interplay, between the genes and environment, that underpins both health, and risk of developing diseases; (iii) understand the causal mechanisms, biomarkers and progression of disease in order to develop new and improved treatments; and (iv) determine genetic risk of metabolic diseases.

We will use the genetic information and wide variety of disease and other phenotypes available in UK Biobank, such as biomarkers and imaging data.

Despite advances in prediction as well as changes in lifestyle and medication these diseases are still on the rise globally. Key challenges in the prevention and management are to: (i) understand better the complex molecular pathways and mechanisms leading to their development, progression, and complications; (ii) determine the causal relation of biomarkers and other putative risk factors, and (ii) find new therapeutic approaches to reduce deaths due to those disorders.

Our proposal will (i) improve understanding of the genetic mechanisms of cardiometabolic and neurological diseases and their comorbidities; (ii) provide a better understanding of how genetics interact with environmental factors in development and progression of these conditions; (iii) and potentially lead to improving their prediction, prevention and treatment.

SUMMARY:
DNA Difference between people may be able to predict if they will develop diseases or respond to medications. However, there is a lack of studies to find these important predictors in people of African Ancestry. We have collected a large cohort of patients with heart problems in which we have collect DNA, how well they respond to medications for their heart problems, and what heart conditions they currently suffer from. We are using this data to find if the predictors in African Ancestry patients are different from what we already know in European patients. This could lead to new diagnostic tests that can be used by doctor to diagnoses and treat patient of African descent.

AIMS:
1) Better understand how genomics and clinical factors influence drug response in African Ancestry patients
2) Better understand how genomics and clinical factors predict risk of disease in African Ancestry patients

DURATION:
We are initially applying for a project duration of 36 months, with a option to extending if studies are ongoing.

PUBLIC HEATH IMPACT:
The studies proposed will generate knowledge in the field of pharmacogenomics (the study of how genetics affect response to drugs) and disease genomics (determining if genetics increases the risk on developing a disease), with the ultimate goal of improving outcomes for African ancestry patients.

This project studies the historical effects of disease epidemics and social stratification on the population of Cambridge through time. Using evidence from ancient and modern DNA as well as methods from archaeology, history, osteoarchaeology, isotopic and osteopathology it asks how healthy and how different from each other were people from different social classes in the past and how they compare to the people from different parts of Britain today, using the UK Biobank data as a reference. One of the main foci of the project is on a recently excavated large sample of urban poor people from the Hospital of St. John (AD 1200-1500), complemented by comparative samples from other medieval social contexts and other historical periods. The results will be analysed both statistically and biographically. A proximate goal is to build a nuanced picture of health, lifestyle and activity in medieval England, one grounded in direct examination of human bodies themselves. The overall goal, however, is to understand the biosocial effects of the Black Death of 1348-1350, an epidemic of bubonic plague which decimated Europe. By comparing samples from before and after the epidemic for a wide range of social and biological indicators, including phenotype related evidence from the UK Biobank, this research will aim to reveal how the plague changed human well-being, activity, mobility, health and the genetic constitution of Europe.

The aim of this project is to validate a blood based risk stratification tool, combining existing and routinely available markers (apolipoprotein-a1, C-reactive protein) with age and gender that could be an additional diagnosis and prognosis tool for screening subjects at high-risk of SARS-Cov-2 infection (SI).

The 2 biomarkers have a published role in the SARS-Cov-2 infection, with normal values depending on age and gender. A score combining the values of the 2 assays, corrected by age and gender, will be correlated to the risk of SI.

The project will last 3 years.

The public health impact could be major, helping the clinicians stratifying the patients according to this novel risk score.

This project investigates ocular and systemic determinants of retinal health and disease risk.
Loss of retinal neurons is a hallmark of ocular ageing and a key contributor to vision loss from diseases such as glaucoma. It may also mirror brain health, offering a non-invasive window into neurodegeneration. However, there is limited large-scale evidence quantifying how retinal nerve fiber layer (RNFL) metrics change across adulthood, and which lifestyle, systemic, and genetic factors accelerate or protect against thinning.
Our goals are to: (1) quantify age-related trajectories of RNFL measures across adulthood, assessing possible non-linearities; (2) identify ocular and systemic risk factors (e.g., blood pressure, metabolic markers, sleep, activity, air pollution exposure, epigenetic clocks) associated with faster RNFL thinning; (3) evaluate genetic contributions using polygenic risk scores and exome/WGS summary metrics; and (4) relate RNFL measures to diagnosed eye and systemic diseases using linked records.

Using the OLINK platform’s inflammation and cardiovascular panels, we have constructed a proteomic age score and calculated proteomic age acceleration (how much proteomic age differs from the chronological age) using data from a Dutch cohort (n=302) established at Radboudumc. However, this and other cohorts that we have access to lack follow-up information. With the UK Biobank data, we aim to understand if the proteomic age acceleration is able to predict cardiovascular events and/or mortality and how it compares to classical risk factors in terms of prediction. We would like to calculate the proteomic age score of the UK Biobank participants for which OLINK measurements are available and to test the predictive power using Cox proportional hazard regression.

Our study aims to develop a new, more accurate way of predicting heart health by using interpretable machine learning to assess biological aging. Unlike chronological age, which only reflects the years a person has lived, biological age can provide insights into the actual health of their cardiovascular system. For this, we will analyze Photoplethysmography (PPG) signals-light-based measurements often used to monitor blood flow-to estimate how a person’s heart and blood vessels are aging.
The goals of this research are threefold: (1) Create an interpretable machine learning model that can predict a patient’s cardiovascular age accurately, (2) Identify and explain the specific biological factors influencing these age predictions, and (3) Investigate the relationship between chronological age and biological age to better understand how they align or differ in the context of heart health.
The scientific rationale behind this study is to bridge the gap between chronological and biological age measurements. Many people with the same chronological age can have vastly different cardiovascular health, which could mean that some individuals are at risk of heart disease sooner than expected. By focusing on biological age, we hope to provide a more personalized assessment of cardiovascular health, leading to better preventative care.
We plan to disseminate out finding through manuscript publication, and the release of our model and code in a publicly accessible repository.

Understanding of the aging process is critical to understanding functional declines and diseases associated with aging such as cardiovascular disease and cancer. In this analysis study, our aim is to improve the understanding of how various features change during the aging process and to construct a mathematical model which can be used to determine how old a person is biologically. This would provide a tool which would allow for investigation of the influence of various factors on the aging process and to flag outliners which have a large deviation of the estimated age and chronological age.
While age estimation has been done and there are many new applications published in the literature this year. However, model calibration with the UK BioBank database is relatively limited to date, and due to its size we expect it to perform well and allow for more advanced model types. Being able to estimate the age of a person allows us to get a sense of their overall health.
In Canada, there are supercomputing resources available for University employed researchers, with greater than 58,000 cores for performing experiments just like this. Such a system is required to run image analysis jobs of this scale and to answer the important questions proposed. A project of this nature is generally tedious to implement as it can take a long to transfer data, arrange and structure the data for such analysis. Furthermore, a project of this nature is often iterative as a processing pipeline of this nature takes time to debug. As I am independent researching working on this project I have a limited time which I can dedicate to the project and currently do not have other team members to allocate to this task. It is for these reasons I am requesting a 3-year timeframe to conduct this project.
The UK BioBank database also has many additional factors such as lifestyle and genetics, which previous age prediction experiments did not have. So, by using these additional variables we will be able to determine how they affect neurodegeneration. In addition to improved age prediction accuracy, we will also glean insights into how these lifestyle and genetic factors change affect the neurodegeneration process.

Around a third of the patients that receive knee replacement still have chronic pain after surgical knee replacement. We have evidence that this may be due to changes to their DNA that could be effective targets for drug development. We plan on comparing the DNA of patients with and without pain to identify the parts of the DNA that could lead us to the most important and relevant changes and lead us to an eventual treatment for those patients that have chronic pain.

People often make their decisions trusting their gut feelings. There is a physiological basis for that. Not only do the gut bacteria and the brain talk to each other, but they also influence each other. Gut microbes affect our mood and behaviour. In turn, the brain can change bacterial composition and function. Such communication between two fundamental systems, which goes in both directions, is known as the gut-microbiota-brain axis. Every day scientific literature offers evidence that disturbances in the relationships between the microbes and the brain have implications for a broad spectrum of health issues. These health conditions include not only various food sensitivities and gut disorders but also psychiatric disorders like depression and brain disorders like autism, Alzheimer’s and Parkinson’s disease. Understanding the influence of the gut-microbiota-brain axis on brain function and behaviour is at the forefront of neuroscience research, and this axis is essential to normal and healthy brain development.

We aim to explore how the relationships between gut microbes and the brain change in ageing and vary across individuals with different cognitive abilities and life experiences. Asking questions on the gut-microbiota-brain axis and its relation to cognition and individual life experience requires highly diverse data from many research fields. The UK Biobank provides a unique opportunity for researchers to access such data. We will look into single-nucleotide polymorphism or SNP, representing variations in single nucleotides at a specific position in the human genome, which previous studies associated with gut microbes. The brain will be described in terms of its functional networks identified by functional MRI. We will consider a wide range of cognitive tests to cluster all the participants into sub-groups, based on the similarity of participants’ cognitive profiles or life experiences such as employment history. The project is extensive in methods and relies on diverse expertise in genomics, neuroimaging, and data science. We expect to get concluding results within two years.

The impact on public health will be reached by identifying bacteria families and functional networks, which are associated with prolonged healthy cognition or risks related to individual life experiences. As we will consider an entire spectrum of ageing trajectories in the gut-microbiota-brain axis, we ultimately target the personalized recommendations for staying within those trajectories that are beneficial for a given sub-group of individuals: promoting the effects of particular bacteria with individual diets or rewiring specific brain networks with personalized rehabilitation.

Age-related hearing loss is the third most common chronic health condition affecting adults in the United States, following hypertension and arthritis and has recently been identified as a risk factor for cognitive decline and dementia. Despite its high prevalence within older adults and its association with dementia, age-related hearing loss remains largely underexplored and unaccounted for in the neuroscience of aging literature. Thus, it is unclear how hearing loss may contribute to brain aging. With this study, we aim to examine how hearing loss impacts brain structure and brain function, and how the APOE-e4 gene (a known risk factor for developing Alzheimer’s disease) may influence these relationships. Our findings could point to novel intervention targets to combat brain aging and Alzheimer’s disease, including the use of hearing aids, which may provide neuroprotective effects against hearing loss.

Our long term goal is to promote healthy ageing in the general population based on new knowledge from individual ageing drivers. Our short term aims are: 1) Characterise DDR ageing signatures and 2) Validate the method in a big cohort of patients from the external databases We hypothesise that individual ageing trajectories can be used to tailor healthy ageing interventions. We strongly believe that implementing effective strategies to reduce disease burden among the elderly will significantly alleviate the strain on the healthcare system. Developing of affordable therapies that promote long-term health will thereby have a huge positive impact on society. the project duration will be 36 months.

Research on predicting human biological aging has been booming during the last few decades. While traditional formula of aging prediction is based on linear models, relatively few works have explored the effectiveness of neural network models, which tends to have the advantage of learning more complex relationship from the data.

In this project, we will study three age-related diseases: cardiovascular disease, cancer, and diabetes. The duration is three years.

Neural networks are computing systems that are inspired by, but not identical to, biological neural networks that constitute human brains. A neural network learns to perform tasks by considering examples with labeled features. For example, in image recognition, they might learn to identify images that contain cats by analyzing example images that have been manually labeled as “cat” or “no cat” and using the results to identify cats in other images. They do this without any prior knowledge of cats because the neural network can learn the characteristics of cat (fur, tails, whiskers, etc.) from a large number of examples.
However, biological gnome data usually consists of hundreds of thousands of features (i.e. genetic markers), while the samples (i.e. patients) are limited due to the convoluted, expensive procedure of gathering data. This introduces the problem of overfitting which leads to poor generalization when applied to different datasets. Overfitting means that the learning is too closely or exactly to a particular set of data, and may therefore fail to fit additional data or predict future observations reliably.

In this project, we will explore several neural network models for handling biological gnome data while taking care of overfitting problems. We propose several models: Basic Neural Network, Dropout Neural Network, Least Absolute Shrinkage and Selection Operator (LASSO) Neural Network, Elastic Net Neural Network, and Correlation Pre-Filtered Neural Network (CPFNN). Our goal is to choose a model with best age prediction by comparing these models and use that model to achieve a decent result on predicting the probability of age-related disease.

The project will have several significant contributions: 1) we will develop a new direction on how to approach high dimension low sample size data in the computer science field; 2) our model can be extended in other fields which also has high dimension low sample size data such as finance; 3) we will provide a sophisticated model on age prediction and age-related disease prediction, which will have a huge impact on clinic fields.

Cardiovascular diseases are among the leading causes of mortality across the world. In particular, coronary artery disease represent a high burden of cardiovascular deaths. The aims of our study is to develop artificial intelligence-based tools using MRI scans, EKG, and patient demographics to predict the risk of adverse outcomes in patients

Type of dataset required: Cardiac imaging data (CMR), patient demographics, outcomes, QOLs, EKGs.

The expected value of research (public interest) is that the development of a non-invasive, novel, AI-aided toolkit will allow accurate prediction of the risk of adverse cardiovascular outcomes. This toolkit will help in future prospective studies to preemptively identify high-risk patients for whom protective strategies may be targeted.

This research aims to develop an AI driven multi disease framework for early detection and risk stratification of chronic diseases common in the elderly such as diabetes, cardiovascular diseases, and neurodegenerative conditions by integrating data from multimodal healthcare sources. By leveraging large-scale biomedical data from the UK Biobank, the research will help identify high-risk individuals and support precision medicine strategies that can improve health outcomes and reduce long-term healthcare costs.

Scientific Rationale:

Chronic diseases in elderly individuals often co-occur, and existing diagnostic pathways are inadequate in addressing comorbidities holistically. Recent advancements in machine learning and IoT-based sensing provide an opportunity to develop predictive models that integrate multi-modal data. The UK Biobank offers a unique, richly annotated dataset ideal for training and validating these models due to its extensive genotypic, phenotypic, and imaging data.

Aims and Objectives:

1. Develop an AI driven healthcare framework for comprehensive prediction of multiple chronic disorders in elderly patients and model their progression over time.
2. Identify key biomarkers and risk patterns to support early intervention and monitoring by integrating genomic, biochemical, lifestyle, and imaging data to enhance prediction accuracy.
3. Optimize AI model for real time deployment and develop a personalized health management system.
4. Compare the model performance against traditional methods and validate the AI-IoT framework to assess its clinical impact.