Machine Learning Reveals COVID-19 Healthcare Financial Vulnerability

This research investigates healthcare financial vulnerability in the United States before and after the COVID-19 pandemic, utilizing Medical Expenditure Panel Survey (MEPS) data from 2019 and 2021. The study defines high financial burden as out-of-pocket healthcare expenditures surpassing 10% of family income and employs both interpretable logistic regression and machine learning models, specifically random forest and gradient boosting, for analysis. The findings reveal that financial vulnerability is strongly linked to poverty status, insurance coverage, and prescription drug spending. While subgroup analyses indicated persistent disparities and some evidence of increased burden among vulnerable populations in 2021, models trained on pre-pandemic data largely retained their predictive power post-pandemic. This suggests that the core predictors of healthcare financial vulnerability remained relatively consistent, highlighting the value of combining statistical modeling with machine learning for population health research.