Recent studies reaffirming geographic disparities in the United States showed that they are not fully explained by socioeconomic differences, suggesting context or environment strongly impacts health. Yet research on environmental influences on health, particularly environmental effect modification of behavioral interventions, has been limited by the cost of computing subject-specific measures of environmental context. We propose to put environmental measures within the reach of non-specialist researchers and health promotion experts by building and validating the Automatic Context Measurement Tool (ACMT). ACMT is a software tool that researchers and practitioners can use to efficiently compile, attribute to individuals, and analyze environmental measures drawn from free and nationally available datasets such as US Census and the National Land Cover Database. After building ACMT, we will take five steps to validate and promote it. First, we will quantify how well these nationally available measures capture health-relevant aspects of study participants' environments by comparing nationally available environment measures predictive of physical activity to locally-available environment measures predictive of physical activity for a cohort based in King County, WA. Second, we will demonstrate how ACMT might apply to multi-site trials by contrasting environment measures predictive of physical activity in King County with those predictive of physical activity in Salt Lake City, UT and Portland, OR. Third, we will use ACMT with large electronic health record (EHR) datasets from Kaiser Permanente Washington (formerly Group Health) patients and UW Medicine patients to explore which environmental measures best predict BMI trajectories in healthy adults. Fourth, we will establish that ACMT can be used to identify environmental modifiers of health intervention effectiveness by comparing environmental predictors of BMI trajectories among adults receiving bariatric surgery compared with obese adults not receiving surgery. Finally, we will ensure ACMT is available on the web with an interface that is usable by non-specialists, recruiting project coordinators from weight management programs and other patient care projects to usability test ACMT and provide feedback allowing us to improve it. Once made publically available, the ACMT will unlock the use of environment measures for researchers and practitioners without geospatial training who had previously been hindered by the considerable expertise (and related expense) required to collect and analyze potential environmental influences on health. The steps we will take to validate ACMT will also provide additional insight into environmental influences on physical activity and obesity. Finally, the complementary training plan comprising coursework, structured mentoring, and experiential learning will let me develop the skills to launch my career as an independent scientist working at the intersection of informatics and epidemiology.
Research on environmental influences on health, particularly environmental effect modification of interventions, has been limited by the cost of computing subject-specific measures of environmental context. We propose to build the Automatic Context Measurement Tool, which will put standardized environmental measures into the hands of non-specialists, unlocking environmental analysis for researchers and practitioners lacking the resources to hire geographic information systems specialists.
