Analyzing Healthcare Coverage in the United States Using Multivariate Machine Learning Models

This study investigates the factors influencing healthcare coverage in the United States using machine learning models. Currently, lack of healthcare coverage is a problem in the United States, with millions uninsured and therefore unable to access essential healthcare services. The research utilizes data from the Behavioral Risk Factor Surveillance System (BRFSS) Survey, focusing on 32 variables related to an individual's health profile. The data was cleaned, normalized, and label encoded before being used to train four different models: XGBoost, Logistic Regression, Random Forest, and AdaBoost. The XGBoost achieved the highest accuracy at 77.91%. The study also employed feature importance and feature magnitude for feature analysis to identify the factors that were most important, finding that employment status, education level, and whether one rents or owns their home were the most consistently important across all models. This research contributes to the understanding of healthcare coverage in the United States and provides insights that could aid policymakers in devising effective strategies. It also highlights the potential of machine learning techniques in analyzing complex datasets and uncovering intricate relationships between variables.