Development of Machine-Learning Models for Predicting Escherichia coli O157:H7 Inactivation on Fresh-Cut Lettuce during Chlorine Washing

Minimally processed leafy greens are a recurrent vehicle for E. coli O157:H7 outbreaks, and chlorine washing remains the primary in-plant hurdle to limit cross-contamination. Yet, the effectiveness of chlorine depends on interacting factors that challenge conventional kinetic models. This study generated a laboratory data set describing E. coli O157:H7 reductions on fresh-cut iceberg lettuce across free-chlorine concentrations of 0-150 mg L^-1 and immersion times of 0-150 s at 4.5 °C. A traditional inactivation model was fitted in MATLAB and compared with three machine-learning (ML) algorithms-support-vector regression (SVR), random-forest regression (RFR), and Gaussian-process regression (GPR)-trained on the same data. Compared to the traditional model (RMSE 0.392; R² 0.721), machine-learning approaches improved predictive accuracy, with GPR (RMSE 0.283; R² 0.852) and SVR (RMSE 0.298; R² 0.836) showing the greatest gains. All ML models maintained stability across validation folds, demonstrating strong generalizability. A user-friendly MATLAB application integrating the fitted models was released as open-source software to facilitate industry and regulatory uptake. These findings confirm the value of data-driven methods for modeling chlorine-wash performance and offer an accessible decision-support tool for optimizing process parameters, thereby helping to reduce the public-health burden associated with leafy-green-related outbreaks.