==Summary==
This umbrella project will integrate UK Biobank’s questionnaires, blood assays, genetics and MRI/CT data to explain why people develop chronic cardiometabolic and neuro-degenerative disease and to build tools that predict-years in advance-who is at highest risk.
We aim to apply for a wide “umbrella” project. However, as applications are required to articulate definable goals, we anchor the umbrella in two exemplar disease domains-cardiometabolic and neuro degenerative-while keeping the analytical workflows modular so they can be re used for additional phenotypes.
==Research questions==
RQ1 – Which modifiable lifestyle, biomarker, genetic and imaging factors causally influence incident type 2 diabetes, coronary heart disease and Alzheimer’s disease?
RQ2 – Can a multimodal model that fuses Tier-1 (clinical), Tier-2 (omics/genetics) and Tier-3 (imaging/WGS) data predict 5- and 10-year risk more accurately than current clinical scores?
RQ3 – Are the causal effects and prediction models transportable to non-UK populations (Israeli national EHRs; NIH All of Us)?
==Objectives==
1. Emulate target trials in UKBB to quantify causal effects of key modifiable exposures.
2. Develop, internally validate and calibrate machine-learning risk models combining clinical, -omics and imaging variables.
3. Externally validate causal estimates and prediction models in Israeli and U.S. cohorts, quantifying transportability.
==Scientific rationale==
UKBB uniquely links deep baseline phenotyping, longitudinal NHS records and large-scale multi-omics/imaging for ~500 000 volunteers. Prior studies usually analyse single diseases or data types, limiting discovery and impact. Our integrated approach pairs modern causal inference with multimodal prediction in a secure enclave, then tests generalisability abroad. The result will be robust causal insights and clinically actionable risk scores that advance precision-prevention.
