A system dynamics model of lead exposure in Newark, New Jersey: Integrating socio-environmental factors and big data for health risk analysis

Abstract

Persistent lead (Pb) exposure in Newark, New Jersey remains a critical urban environmental health issue shaped by socioeconomic, environmental, and behavioral determinants. Modeling these complex relationships over time poses significant challenges. In this study, we developed and validated a system dynamics exposure model that integrates environmental data, socioeconomic variables, and a Google Trends-based public awareness index to simulate elevated blood lead levels (EBLL) and explore policy-relevant scenarios in the Newark context. The model was constructed in Stella Architect and calibrated to Newark data from 2010 to 2020. Primary inputs included industrial Pb releases, housing age, poverty, health insurance, and a Google Trends-based public awareness index. Regression-based covariation mining (R² ≥ 0.5) informed feedback pathways, while model behavior was evaluated using RMSE, initial-stock sensitivity testing, data-precision testing, boundary checks, and temporal comparison with observed EBLL trends. The model closely reproduced observed EBLL trajectories for adults and children (adult RMSE = 0.05332; child RMSE = 0.02917) and showed a strong covariation between the awareness index and EBLL (R² = 0.767). Key covarying factors include socioeconomic vulnerability and public awareness dynamics. The business-as-usual run is interpreted as conditional scenario exploration rather than deterministic prediction, suggesting a possible near-term EBLL increase before gradual decline if historical relationships persist. This framework advances exposure science by combining dynamic modeling with social sensing data while explicitly bounding causal interpretation.