My name is Alana Schreibman, and I am a rising junior at the University of Pennsylvania from Sarasota, Florida. I am majoring in Earth Science and minoring in Chemistry and Chinese Studies. My academic interests are at the intersection of human health, environmental science, and social justice.
What is your summer research project?
This summer, I worked in the Himes Lab under Dr. Blanca Himes on predicting geospatial adult asthma exacerbation risk in Philadelphia using electronic health record (EHR) data, with a focus on comparing EPA-derived ground-level pollution to remotely-sensed pollution. This project required leveraging large amounts of data to model exacerbation risk for a cohort of asthma patients. We set exacerbation counts over the study period as the outcome of interest, and used two approaches: a proportional odds model and a generalized additive model with two-dimensional smoothing based on latitude and longitude from de-identified patient addresses. EHR-derived data was from University of Pennsylvania Hospital System 2017-2019 patient encounters. We also integrated socioeconomic variables, namely Area Deprivation Index (ADI) and three nitrogen dioxide (NO2) measures: in situ EPA AirData, TROPOMI satellite tropospheric column density, and satellite-derived ground estimates.
What are the implications of your research?
Philadelphia asthma prevalence rates are consistently higher than national rates, and exacerbations are still large contributors to asthma-related mortality. NO2 (independently and as part of a pollutant mixture) and socioeconomic disadvantage have been associated with asthma exacerbations. Past work has demonstrated that these variables are spatially heterogeneous across Philadelphia; therefore, we aimed to understand their correlation with exacerbations.
Our choice to use EHR-derived data and multiple different NO2 measurements are both important to the implications of our research. EHR is a valuable alternative to traditional epidemiologic research that has limitations, but captures large, diverse cohorts, including vulnerable populations. These cohorts live in geographically diverse areas, allowing for a strong assessment of spatial risk. Additionally, EPA pollutant monitors are sparse and cannot completely capture variation within urban areas. Accordingly, we incorporated NO2 data derived from TROPOMI, which has a very fine spatial resolution. In short, we aimed to use these methods for a fine-scale assessment of spatial asthma risk across Philadelphia. We hypothesized that this is important for understanding factors underlying the city’s health disparities and for evaluating the utility of our methods for future research.
What new skills have you gained through your research?
This project has deepened my understanding of working with large data sets, including benefits, inherent biases, and best practices for analysis. Specifically, understanding how to clean and analyze EHR-derived data is a skill that I will use frequently in future research. I also improved my skills in R and working in high-performance computing environments, and gained a better understanding of biomedical and environmental informatics literature.
