eng Science and Education Publishing American Journal of Epidemiology and Infectious Disease 2333-1275 2025-05-08 13 1 10 18 10.12691/ajeid-13-1-2 AJEID20251312 article Apeksha Mewani abcdef@gmail.com 1 Vincent Jones II 2 Alejandro Sanchez 3 Department of Health Equity, Administration, and Technology, Lehman College, CUNY Department of Health and Human Performance, York College, CUNY, USA School of Medicine, American University of the Caribbean, St. Marteen This study determines the best machine learning (ML) models to predict the most accurate results in COVID-19 seropositivity using existing data. The study used the New York City Community Health Survey (NYC CHS) 2020 dataset for the analysis, and a predictive modeling approach to develop and select an optimal ML model that would accurately predict COVID-19 seropositivity from various ML algorithms. Thus, the LightGBM was found to have the highest Area Under the Curve and overall test metrics including accuracy, precision, and recall, and was therefore selected as the best-performing machine learning model. https://pubs.sciepub.com/ajeid/13/1/2/ajeid-13-1-2.pdf COVID-19 Machine learning emergency preparedness predictive modeling seropositivity health education