Authors: Jiue-An Yang*, University of California - San Diego, Marta Jankowska, University of California - San Diego, Chad Spoon, University of California - San Diego, Steven Zamora, University of California - San Diego, Tarik Benmarhnia, University of California - San Diego
Topics: Health and Medical, Spatial Analysis & Modeling
Keywords: Environmental Exposure, Health Outcome, Mobility, Spatial-Temporal Analysis
Session Type: Virtual Paper
Start / End Time: 9:35 AM / 10:50 AM
Room: Virtual 8
Presentation File: No File Uploaded
Quantifying exposure is an essential step in the workflow of understanding the impact of environmental exposure on personal and individual health outcomes. The ability to capture continuous location using GPS-enabled devices has driven the paradigm shift in this domain, moving from static, neighborhood-based exposure to dynamic exposure based on the mobility pattern. However, challenges remain at how dynamic exposure can be better measured to reflect the actual human-environment interactions during the daily life-space of individuals. In this paper, we report our work in comparing three methods for measuring exposure that account for both spatial and temporal aspects of mobility data: Kernel Density Estimation (KDE), spatial joining, and Density Ranking (DR). Each method is evaluated with different types of environmental layers including point (fast food stores), line (bike paths), area (parks and recreation facilities), and continuous surface (PM2.5 pollution) layers. As an example, we use the minute-level GPS location from 602 participants with an average of eleven days of data to compare the methods. Location information is enriched with accelerometer data to further derive modes of physical behaviors at the moment including sedentary, walking, and in-vehicle. We will walk through the workflow of data preparation and computation for each method. Summary statistics for participant exposures and outcome interpretability will be discussed as well.