You are here:
Integrating Time-Activity and Air Sensors into Smartphone-based PM2.5 and Ozone Exposure Model for Epidemiologic Studies
Citation:
Breen, M., V. Isakov, C. Skarke, C. Seppanen, AND S. Arunachalam. Integrating Time-Activity and Air Sensors into Smartphone-based PM2.5 and Ozone Exposure Model for Epidemiologic Studies. International Society for Environmental Epidemiology, Athens, GREECE, September 18 - 21, 2022.
Impact/Purpose:
Epidemiologic studies of fine particulate matter (PM2.5) and ozone (O3) often use outdoor concentrations as exposure surrogates, which can induce measurement error. The goal of this study was to improve PM2.5 and O3 exposure assessments for a repeated measurement study with 20 individuals in Philadelphia, Pennsylvania by applying a smartphone-based exposure model called TracMyAir. Our study demonstrates the capability of extending TracMyAir with air pollutant and time-activity sensors and models to estimate individual-level PM2.5 and O3 exposure metrics for an epidemiologic study, in support of improving risk estimation.
Description:
BACKGROUND AND AIM: Epidemiologic studies of fine particulate matter (PM2.5) and ozone (O3) often use outdoor concentrations as exposure surrogates, which can induce measurement error. The goal of this study was to improve PM2.5 and O3 exposure assessments for a repeated measurement study with 20 individuals in Philadelphia, Pennsylvania by applying a smartphone-based exposure model called TracMyAir. METHODS: We developed TracMyAir, which is a smartphone (iPhone,Android) application that determines multiple tiers of individual-level exposure metrics in real-time for ambient and non-ambient PM2.5 and O3 using outdoor concentrations, home building characteristics, weather, time-locations, and time-activities. In this study, we extended TracMyAir by including (1) outdoor concentrations from an air quality model (CMAQ), (2) indoor and outdoor PM2.5 concentrations from low-cost air sensors (PurpleAir) and a building infiltration model, (3) a microenvironment model (MicroTrac) based on time-resolved smartphone geolocations, and (4) inhaled ventilation models based on physical activity data from smartphone and smartwatch accelerometers and heart rate sensors. The five tiers of exposure metrics with increasing information needs and model complexity include: residential air exchange rates (AER, Tier 1), infiltration factors (Finf_home, Tier 2), indoor concentrations (Cin_home, Tier 3), exposures (E, Tier 4), and inhaled doses (D, Tier 5). We applied TracMyAir to determine hourly PM2.5 and O3 exposure metrics for three consecutive days for 20 participants consisting of two age groups (18-30 years old, n=10; 55-75 years old, n=10). RESULTS: The TracMyAir predictions showed considerable temporal and house-to-house variability of AER, Finf_home, and Cin_home (Tiers 1-3), and person-to-person variability of E and D (Tiers 4-5). CONCLUSIONS: Our study demonstrates the capability of extending TracMyAir with air pollutant and time-activity sensors and models to estimate individual-level PM2.5 and O3 exposure metrics for an epidemiologic study, in support of improving risk estimation.