Data mining epidemiological relationships

Everyday lifestyle choices—like diet, exercise, or smoking—can raise or lower a person's risk of common diseases, but pinning down which factors actually cause harm, rather than just being correlated with it, has been notoriously difficult. This research programme is building automated data mining tools that use genetic information to separate true causal risk factors from misleading associations. The team is also constructing a "knowledge graph"—a vast, interconnected map of biomedical evidence—that links genetic data, disease mechanisms, and drug effects. This allows them to systematically search for new drug targets, predict side effects, and identify existing drugs that could be repurposed for other conditions. If successful, the tools and knowledge graph will be made openly available to the global research community. This could accelerate the discovery of modifiable risk factors for diseases like heart disease or diabetes, and speed up the identification of new treatments—without requiring new clinical trials for every candidate. The work is primarily methodological and fundamental in nature, but similar open-source analytical tools have previously transformed how researchers use large-scale population data to improve public health.