High Resolution and Spatiotemporal Place-Based Computable Exposures at Scale

Abstract

Place-based exposures, termed "geomarkers", are powerful determinants of health but are often understudied because of a lack of open data and integration tools. Existing DeGAUSS (Decentralized Geomarker Assessment for Multisite Studies) software has been successfully implemented in multi-site studies, ensuring reproducibility and protection of health information. However, DeGAUSS relies on transporting geomarker data, which is not feasible for high-resolution spatiotemporal data too large to store locally or download over the internet. We expanded the DeGAUSS framework for high-resolution spatiotemporal geomarkers. Our approach stores data subsets based on coarsened location and year in an online repository, and appropriate subsets are downloaded to complete exposure assessment locally using exact date and location. We created and validated two free and open-source DeGAUSS containers for estimation of high-resolution, daily ambient air pollutant exposures, transforming published exposure assessment models into computable exposures for geomarker assessment at scale.

Figure 1.

An overview of the DeGAUSS approach used for large spatiotemporal data. Input start and end dates are expanded into a daily time series for each row in the input data. The calendar year and a coarsened geohash are used to download a spatiotemporal subset of exposure estimates. The files are cached locally, and exposure estimates are merged in based on exact date and location of the input data. Spatiotemporal pseudo-identifiers are removed by the user and exposures can be optionally averaged over specific time periods to reduce the risk of reidentification if sharing data.

Figure 2.

(Left) S2 and H3 geohashes covering the contiguous United States colored by estimated population. (Right) “Safe Harbor” geohashes aggregated to prevent any single geohash from containing less than 20,000 estimated individuals. Aggregated geohashes are outlined in red.

Figure 3.

System Usability Survey Results. In the four listed questions, respondents summarized their experience using DeGAUSS.