Chronic diseases such as type 2 diabetes (T2DM) arise from complex interactions among genetic, behavioral, and environmental factors. While polygenic risk scores (PRS) have been widely adopted to quantify inherited risk, conventional approaches often aggregate genetic effects into a single score, limiting insight into the underlying biology of disease susceptibility. To address this, we present a platform that integrates partitioned polygenic risk scores (pPGS) —which reflect distinct physiological domains such as β-cell function, insulin processing, adipose tissue biology, and hepatic metabolism — with longitudinal lifestyle data and environmental context.

At the core of this approach is the OneHealth Knowledge Base, a structured repository of directional relationships between lifestyle factors (e.g., nutrients, foods, physical activity patterns, social behaviors) and specific health outcomes. These relationships were extracted from over 40 million scientific articles in PubMed and PMC using AI methods enhanced with manual curation. The knowledge base is being expanded to include relationships between lifestyle factors and molecular targets involved in disease development.

These relationships are operationalized through a digital healthware platform that includes: (1) assessments of individual-level behaviors and exposures; (2) linkage to community-level determinants of health using public datasets such as the American Community Survey (ACS), NHANES, and the U.S. Bureau of Labor Statistics; and (3) integration with genomic data via pPGS.

This presentation will describe the methodological framework used to develop and validate the partitioned scores, the annotation and inference pipeline behind the knowledge base, and the design of digital tools for capturing lifestyle behavior in free-living individuals. We will illustrate how the integration of pPGS with lifestyle data and curated evidence can be used to generate mechanistically-informed hypotheses, stratify risk, and guide the design of personalized interventions. Implications for research, clinical utility, and public health implementation will also be discussed.