This study investigates the association between social and environmental exposures and chronic disease burden in the US, using the Environmental Justice Index (EJI) and explainable AI (XAI) methods. The EJI dataset contains social vulnerability, environmental burden, and health vulnerability indicators for over 71,000 census tracts. Using both traditional statistical approaches (Spearman correlation and linear regression) and machine learning (ML) models (Random Forests and Neural Networks), this study assessed the association between social/environmental factors and health indicators. Our results show that ML outperformed linear regression in model fitting, achieving 82–91% in R2. This indicates that, to a very large extent, variance in disease prevalence can be explained by social/environmental factors, and their relationships are non-linear. The XAI-generated impact scores further identified individual factors positively and negatively associated with specific diseases, which can inform publica health policies and interventions.