Refining source-specific lung cancer risk assessment from PM2.5-bound PAHs: Integrating component-based potency factors and machine learning in Ningbo, China

The component-based potency factor approach, combined with benzo[a]pyrene (BaP) unit risk values from the World Health Organization (WHO), is commonly used to assess lung excess cancer risk (LECR) from polycyclic aromatic hydrocarbons (PAHs). However, this method may overestimate LECR, particularly when highly carcinogenic PAHs are included. In this study, we employed BaP unit risk values from both the WHO and the Environmental Protection Agency (EPA) to estimate LECR in Ningbo, China, revealing that incorporating high-carcinogenic PAHs into the component-based potency factor approach, along with WHO unit risk factors, leads to an overestimation of LECR by more than tenfold. We identified a moderate PAH exposure risk level (>1.0 ×10⁻⁶) in Ningbo and used advanced machine learning (ML) algorithms, random forest (RF), extremely randomized trees (ERT), and extreme gradient boosting (XGBoost), to improve the accuracy of source-specific LECR assessments. ERT emerged as the most robust algorithm, identifying industrial emissions, coal combustion, and gasoline engine exhaust as the primary contributors to elevated LECR in Ningbo. This study underscores the need for precise, source-specific LECR estimation to effectively mitigate PAH pollution and reduce lung cancer risks. By integrating ML techniques into risk assessment methodologies, we provide a robust framework for global application, enhancing public health protection. Our findings also highlight the importance of refining risk evaluation strategies and pave the way for future research to validate and adapt these models in diverse environmental settings.