Identifying Potential Factors Associated With Racial Disparities in COVID-19 Outcomes: Retrospective Cohort Study Using Machine Learning on Real-World Data

• DOI: 10.2196/54421
• Publication Date: 2024-09-26T19:00:35Z
• Authors: Osama Dasa; Chen Bai; Ruba Sajdeya; Stephen E Kimmel; Carl J Pepine; Matthew J Gurka J; Reinhard Laubenba...; Thomas A Pearson; Mamoun T Mardini
• Keywords: Male; Adult; Original Paper; Adolescent; COVID-19; Health Status Disparities; Middle Aged; White People; Machine Learning; Black or African American; Cohort Studies; Young Adult; Socioeconomic Factors; Risk Factors; Florida; Humans; Female; Public aspects of medicine; RA1-1270; Retrospective Studies; Aged

Background Racial disparities in COVID-19 incidence and outcomes have been widely reported. Non-Hispanic Black patients endured worse disproportionately compared with non-Hispanic White patients, but the epidemiological basis for these observations was complex multifaceted. Objective This study aimed to elucidate potential reasons behind of experienced by how variables interact using an explainable machine learning approach. Methods In this retrospective cohort study, we examined 28,943 laboratory-confirmed cases from OneFlorida Research Consortium’s data trust health care recipients Florida through April 28, 2021. We assessed prevalence pre-existing comorbid conditions, geo-socioeconomic factors, structured electronic records cases. The primary outcome a composite hospitalization, intensive unit admission, mortality at index admission. developed validated model Extreme Gradient Boosting evaluate predictors rank them importance. Results Compared Blacks were younger, more likely be uninsured, had higher emergency department inpatient visits, regions area deprivation rankings pollutant concentrations. highest burden comorbidities rates outcome. Age key predictor all models, ranking patients. However, congestive heart failure predictor. Other variables, such as food environment measures air pollution indicators, also ranked high. By consolidating into Elixhauser Comorbidity Index, became top predictor, providing comprehensive risk measure. Conclusions reveals that individual factors significantly influence COVID-19. It highlights varying profiles among different racial groups. While findings suggest disparities, further causal inference statistical testing are needed fully substantiate observations. Recognizing relationships is vital creating effective, tailored interventions reduce enhance across socioeconomic