Predictive Model for Listeria monocytogenes in RTE Meats Using Exclusive Food Matrix Data

Post-processing contamination of Listeria monocytogenes has remained a major concern for the safety of ready-to-eat (RTE) meat products that are not reheated before consumption. Mathematical models are rapid and cost-effective tools to predict pathogen behavior, product shelf life, and safety. The objective of this study was to develop and validate a comprehensive model to predict the Listeria growth rate in RTE meat products as a function of temperature, pH, water activity, nitrite, acetic, lactic, and propionic acids. The Listeria growth data in RTE food matrices, including RTE beef, pork, and poultry products (731 data sets), were collected from the literature and databases like ComBase. The growth parameters were estimated using the logistic-with-delay primary model. The good-quality growth rate data (n = 596, R² > 0.9) were randomly divided into 80% training (n = 480) and 20% testing (n = 116) datasets. The training growth rates were used to develop a secondary gamma model, followed by validation in testing data. The growth model's performance was evaluated by comparing the predicted and observed growth rates. The goodness-of-fit parameter of the secondary model includes R² of 0.86 and RMSE of 0.06 (μ_max) during the development stage. During validation, the gamma model with interaction included an RMSE of 0.074 (μ_max), bias, and accuracy factor of 0.95 and 1.50, respectively. Overall, about 81.03% of the relative errors (RE) of the model's predictions were within the acceptable simulation zone (RE ± 0.5 log CFU/h). In lag time model validation, predictions were 7% fail-dangerously biased, and the accuracy factor of 2.23 indicated that the lag time prediction is challenging. The model may be used to quantify the Listeria growth in naturally contaminated RTE meats. This model may be helpful in formulations, shelf-life assessment, and decision-making for the safety of RTE meat products.