Asthmatics commonly visit emergency departments for various reasons, and there is an apparent disparity in the reason, timing, and severity of these visits based on race. A study published in the Journal of Asthma attempted to use machine learning to conduct a thorough retrospective analysis of health records in order to determine the relevant variables that determine the disparate outcomes that were found.
Data was examined from a total of 42,375 patients. Of these, 14,491 were of African American ancestry, while 27,884 were of European American ancestry, according to the relevant electronic health records. The data was stratified according to demographic, clinical and environmental factors, including age, race, gender, smoking status, and pollen and mold exposure.
A variety of machine learning techniques were used to build predictive models related to expected emergency department visitation rates. These include extreme gradient boosting, decision trees, and random forest methods. Both race-specific and season-specific models were created on the basis of the various types of patient data that were available to the researchers.
Significant disparities were found between African American and European American patients, especially as regards emergency department visits and Forced Expiratory Volume (FEV1) percentage rates. Extreme gradient boosting, in particular, was able to accurately classify emergency department visitors based on race.
This points to key predictors that need to be targeted in an attempt to lessen disparities and improve overall public health conditions for asthma patients. In particular, socio-economic status and its relationship to particulate matter and mold require targeting in order to improve the outcomes of non-white asthma patients.
Unraveling racial disparities in asthma emergency department visits using electronic healthcare records and machine learning. 2020. Adeboye A. Adejare, Yadu Gautam, Juliana Madzia, Tesfaye B. Mersha10.1080/02770903.2020.1838539. Journal of Asthma