Quantitative Microbial Risk Assessment of E. coli in Riverine and Deltaic Waters of Northeastern Greece: Monte Carlo Simulation and Predictive Perspectives

This study presents a comprehensive Quantitative Microbial Risk Assessment (QMRA) for <i>Escherichia coli</i> in northeastern Greece's riverine and deltaic aquatic systems, evaluating potential human health risks from recreational water exposure. The analysis integrates seasonal microbiological monitoring data-<i>E. coli</i>, total coliforms, enterococci, <i>Salmonella</i> spp., <i>Clostridium perfringens</i> (spores and vegetative forms), and physicochemical parameters (e.g., pH, temperature, BOD₅)-across multiple sites. A beta-Poisson dose-response model within a Monte Carlo simulation framework (10,000 iterations) was applied to five exposure scenarios, simulating varying ingestion volumes for different population groups. Median annual infection risks ranged from negligible to high, with several locations (e.g., Mandra River, Konsynthos South, and Delta Evros) surpassing the World Health Organization (WHO)'s benchmark of 10^-4 infections per person per year. A Gradient Boosting Regressor (GBR) model was developed to enhance predictive capacity, demonstrating superior accuracy metrics. Permutation Importance analysis identified enterococci, total coliforms, BOD₅, temperature, pH, and seasons as critical predictors of <i>E. coli</i> concentrations. Additionally, sensitivity analysis highlighted the dominant role of ingestion volume and <i>E. coli</i> levels across all scenarios and sites. These findings support the integration of ML-based tools and probabilistic modelling in water quality risk governance, enabling proactive public health strategies in vulnerable or high-use recreational zones.