Comprehensive assessment of E. coli dynamics in river water using advanced machine learning and explainable AI

The discharge of untreated municipal wastewater has resulted in faecal contamination of river water, posing severe public health risks, and has challenged safe irrigation. Therefore, the present study quantified the Escherichia coli (E. coli) contamination in three rivers of the Tripura region and assessed the impact of land use (LU) patterns on E. coli dynamics using spatial distribution maps. Further, the Quantitative Microbial Risk Assessment (QMRA) model is utilized to evaluate microbial risks associated with farmers using contaminated river water for irrigation. Finally, this study is the first of its kind to use and compare three hyper-tuning frameworks, which included Bayesian optimization, Tree-based Pipeline Optimization Tool, and Optuna, to predict E. coli concentration. This work also utilizes the Explainable AI (XAI) based Shapley Additive Explanations (SHAP) and Local Interpretable Model-Agnostic Explanations (LIME) for global and local site-specific sensitivity analyses, providing interpretable and actionable insights. The findings show that water quality in all three rivers is unsuitable for drinking primarily due to elevated E. coli levels. Stable pH levels and favorable temperatures support E. coli growth, intensifying the contamination risk. The QMRA model further indicates a 0.01- 0.57 probability of significant health risks for farmers using contaminated water. Additionally, the machine learning approaches, along with statistical metrics and cumulative density function plots, reveal the superior performance of the Optuna-optimized extreme gradient-boosting (XGBoost) model over the random forest (RF) and gradient-boosting machine models (GBM). XAI recognized electrical conductivity and total dissolved solids as the most influential factors affecting the E. coli concentrations. Overall, this framework can predict regions impacted by faecal contamination, supporting the sustainable development goals for clean water and health.