Predictive modeling of PFAS behavior and degradation in novel treatment scenarios: A review

Abstract

Per- and polyfluoroalkyl substances (PFAS) are persistent environmental contaminants that resist conventional water treatment methods, raising concerns about their impact on human health and ecosystems. As PFAS contamination becomes increasingly widespread, the need for novel, effective treatment solutions have grown. Predictive modeling offers a promising approach to evaluate PFAS behavior, removal efficiency, and transformation pathways in emerging treatment technologies. This narrative review explores current advancements in predictive models for PFAS remediation, focusing on methods that incorporate PFAS structural characteristics, environmental factors, and treatment type. Three main modeling approaches are discussed: empirical, mechanistic, and machine learning models, each with unique strengths and limitations depending on data availability and treatment conditions. The review also addresses recent developments in advanced treatment systems such as advanced oxidation processes (AOPs), electrochemical treatment, and adsorption, as well as the role of machine learning in optimizing treatment predictions. Key challenges, including data limitations, transformation product toxicity, and model validation, are examined, with recommendations for future research emphasizing data expansion, integration of toxicity predictions, and enhanced model interpretability. By tailoring predictive models to PFAS-specific variables and diverse treatment conditions, researchers can advance sustainable PFAS management practices and guide effective remediation strategies for contaminated sites.