Human health emerges from interactions among genetics, molecular biology, lifestyle and the environment. This project asks: how do host genetics, immune status, molecular profiles, lifestyle and environmental exposures jointly determine susceptibility, severity and recovery for both chronic and infectious diseases?
Using UK Biobank’s multimodal resources (genotypes, proteomics/metabolomics, multi-organ imaging, serology, electronic health records, lifestyle questionnaires and longitudinal follow-up), we will develop integrative, generalizable models to (1) predict individual risk and prognosis, (2) identify biomarkers for early detection and prevention, and (3) enable precision interventions.
UK Biobank’s scale and deep phenotyping uniquely permit joint analysis of host genetics, immune and molecular signatures, pathogen exposure (serology/vaccine records) and exposomic variables. Integrating these layers can reveal causal pathways linking exposures and biology to outcomes, improve risk stratification, and uncover targets for intervention.
Approach (summary): apply multimodal data fusion and machine learning (including representation learning and transformer-based models), causal inference (e.g., Mendelian randomization, structural models), mixed-effects longitudinal modeling, and exposome analysis. For infectious diseases we will explicitly model serological evidence, vaccine history and immune biomarkers alongside host and environmental factors to predict infection risk, severity and long-term sequelae.
Impact and dissemination: outputs will include validated predictive models, candidate biomarkers, and reproducible code; results will be shared via open-access publications, deposited models/tools, and accessible summaries for stakeholders. This work aims to advance predictive, preventive and precision medicine by translating multimodal insights from UK Biobank into actionable public-health and clinical strategies.
