Interrogating the multi-omic architecture of the exposome and intervention from populations to individuals

The “exposome”—the array of environmental exposures from diet to chemical to infection—is vast, dynamic, and intertwined with biology across scales, but unlike the genome remains elusive despite thousands of candidate studies. This talk traces a practical pipeline for scaling exposome-phenome associations and calibrating them with intervention. First, we show population-scale maps relating hundreds of environmental and lifestyle factors to diverse phenotypes, quantifying realistic effect sizes, replication, and variance explained. Modeling correlated exposures jointly reveals modest but meaningful gains—an architecture reminiscent of polygenic traits—and sets baselines for discovery, prioritization, and study design. Second, we move to a promising future, integrate multi-omic layers—especially proteomics and metabolomics—to characterize trajectories of metabolic dysfunction and to nominate biology-anchored targets. By leveraging observational cohorts alongside interventional datasets (e.g., GLP-1–based therapy), we identify response-linked signatures for experimental opportunities. Third, we show recent work in the group to untangle intra-individual variation in glucose response, using AI approaches such as interpretable state-space models that fuse continuous glucose monitoring with wearable signals to forecast short-term risk and run counterfactual “what-if” scenarios for personalized self-management. We will also discuss emerging consortium for exposomic research, Nexus-exposomics.org.