Human health and disease arise from a complex interplay of genetic, biological, lifestyle, and environmental factors. Traditional studies have often focused on single data modalities, limiting their ability to capture this complexity.
At EIT, we aim to develop a broad range of models using UK Biobank’s large multimodal dataset. The models will encompass genomics, imaging, biomolecular, health record, lifestyle, and longitudinal study data. Leveraging this comprehensive dataset, we will combine emerging analytical techniques in the fields of machine learning, biostatistics, and epidemiology (such as multimodal transformers, mixed-effects models, causal inference, exposome analysis) to build large-scale, multi-purpose models capable of uncovering subtle biological patterns that are otherwise inaccessible through conventional approaches, such as unimodal analysis.
The project’s objectives include:
– Improve the understanding of complex disease mechanisms across genetic, molecular, and environmental dimensions;
– Discover biomarkers for early detection, monitoring, and prevention of complex diseases and demonstrating the applications thereof;
– Accelerate research in pathogen biology, immunology, and synthetic biology;
– Improve overall public health and therapeutics enabled by more efficient drug discovery and administration.
Objectives for our models culminate in the overall benefit to the public by improving public and scientific understanding of complex diseases, accelerating effective therapeutic development, thus alleviating the burden on the health system across the UK and globally.
Findings will be shared through accessible summaries and with the wider scientific community through peer-reviewed publications, open access resources, and public presentations to ensure broader impact.