Presenter: Adriana Payan-Medina, School of Medicine, Biomedical Informatics
Authors: Adriana Payan-Medina, Naomi Riches PhD, MSPH, Ramkiran Gouripeddi MBBS,MS, Julio C. Facelli, PhD
Faculty Advisors: Ramkiran Gouripeddi, School of Medicine, Biomedical Informatics
Institution: University of Utah
Air pollution has posed an exacerbating threat to Utah citizens as motor vehicle emissions, fossil fuel contributions, and Winter inversions jeopardize human health. Chemical pollution can be inhaled deep into the lungs, contributing to lung damage before crossing the alveolocapillary membrane to enter circulation within the bloodstream, which can lead to tissue damage, and increased cardiovascular or respiratory health risks, and cause inflammatory or metabolic insults. These conditions, stimulated by adverse air quality (AQ), have promoted the increase of coronavirus death rates, demonstrating the urgency of widespread and accurate pollutant measurements. Air quality data is obtained primarily from ground-based air quality monitors which obtain measurements of the greatest accuracy when patients live within 25 miles of the monitor, and due to geographical and financial limitations, monitors are generally located in population-dense areas. Contrarily, satellite data from the National Aeronautics and Space Administration (NASA) gives access to spatiotemporal air pollution data through daily global exposure coverage of NO2 and PM2.5 concentrations through geographic coordinate-specific values. Through NASA’s MERRA PM2.5, NASA's Ozone Monitoring Instrument NO2, and the Environmental Protection Agency’s NO2 and PM2.5 concentration measurements, 15 years of chemical pollution datasets specific to each United States county were compiled. Through Python’s pandas package, a descriptive statistical analysis was made for overlapping monitor and satellite locations and dates. We will use Python’s scikit-learn package to perform unsupervised machine learning to provide a quantitative evaluation of the precision of chemical pollutant data from satellites compared to chemical pollutant data from ground-based monitors. With the results of this analysis, we hypothesize that we will attain a more thorough representation of air pollution exposure to a wider swath of the US population, and the subsequent evaluation of the impacts atmospheric pollution has on various diseases such as COVID-19 and Type 2 Diabetes Mellitus.