American Association for Aerosol Research - Abstract Submission

AAAR 36th Annual Conference
October 16 - October 20, 2017
Raleigh Convention Center
Raleigh, North Carolina, USA

Abstract View


Assessing Exposure Misclassification Error Using Cell Phone Location Data

Haofei Yu, ARMISTEAD G RUSSELL, James Mulholland, Georgia Tech

     Abstract Number: 499
     Working Group: Aerosol Exposure

Abstract
In air pollution epidemiologic studies, individual subject’s exposure to particulate matter is usually modeled as that occurring at their corresponding home address. When spatiotemporal individual mobility is not accounted for, exposure misclassification errors are likely. In this study, we applied a detailed cell phone location dataset, the call detail record (CDR), to assess the potential exposure misclassification errors in the home-based approach, and to demonstrate its potential in improving exposure estimates of individuals to PM, PM species and other air pollutants. The CDR database was collected from Shenzhen, China on a mid-week day in October, 2013. It contains 9,886 unique simcard IDs and approximately 4.6 million location records. Air pollution exposures were calculated for each ID by matching their detailed location data with hourly ambient concentrations of six chosen pollutants, which were modeled by the Community Multi-scale Air Quality model (3 km resolution) and fused with observational data. The estimated exposures were compared with those obtained using the home-based approach. Our results show substantial differences between exposures estimated by the CDR-based and home address-based approaches, and indicate likely exposure misclassification errors when spatiotemporal subject mobility is not accounted for. The home address-based approach tends to over-estimate exposures for individuals with higher exposure levels, and under-estimate exposures for those with lower exposure levels. Our findings show potential of the cell phone location-based approach for improving exposure estimates in air pollution epidemiology and health impact studies.