Get started
Work

An OMOP-Compatible Exposure Infrastructure for GeoAI-Enabled Environmental Health Research: The OHDSI GIS Toolchain

Year: 2026
Type: preprint
Abstract: Environmental and socio-economic contexts shape population health. Despite increasing availability of data about such contexts, no open infrastructure standardizes these data into person-place-time exposure records for federated health research. We present the OHDSI GIS toolchain, an open-source infrastructure integrating geospatial data with the OMOP Common Data Model through a FAIR catalog, a GIS vocabulary, local geocoding, and a geometry-attribute index that registers every source uniformly. We tested the architecture on a synthetic cohort of 40,000 persons in real US census tracts, with eight simulated PM2.5 effects and six null outcomes. Derived exposures matched an independent answer key, and known effects were recovered. Displacing residences by empirical geocoding error changed estimates little for a smooth regional exposure but attenuated them by up to half for exposures varying over 100 m. These results show that the toolchain and its standards enable reproducible, transparent linkage of geospatial sources with health records.
DOI: 10.21203/rs.3.rs-11436369/v1
Authors: Andrew E Williams, Kyle Zollo-Venecek, Robert Miller, Polina Talapova, Timothy Norris, Jay Greenfield, Nitin Parikh, Jimmy Phuong, Anne E Thessen, Timothy D Rossi, Alexia Sampri, Jiawen Liao, Sara Khalid, Maksym Trofymenko, Jared Houghtaling
Institutions: Tufts Clinical and Translational Science Institute, University of Miami, Committee on Data of the International Science Council, University of Washington, University of North Carolina at Chapel Hill, Cleveland Clinic, University of Oxford
Language: English
Open Access status: green
OpenAlex ID: W7225743922
Cites: 25
Cited by:
Related to: