Integrating kinetic models and BP-ANN for predicting growth and shelf-life of Listeria monocytogenes in ready-to-eat salads

Abstract

Ready-to-eat (RTE) salads are popular for their convenience and health benefits. However, their reliance on refrigeration increases the risk of contamination by Listeria monocytogenes. In this study, vegetable salads (VS) and salads with 10% (10S), 20% (20S), and 30% (30S) chicken breast were inoculated with a L. monocytogenes cocktail and stored at 4, 10, or 15 °C for 14 days. The results indicated that higher storage temperatures and longer durations significantly increased microbial activity. A positive relationship was observed between chicken breast content and total viable count (TVC) as well as L. monocytogenes growth. At the end of storage, the TVC in 30S was 8.57 ± 1.39 log CFU/g higher than that in VS. As determined by the logistic model, the highest maximum growth rate (μ_max) observed in 30S at 15 °C was 0.0589 ± 0.0004 log CFU·g⁻¹·h⁻¹. Salads with meat had a shelf life of up to 9 days, while the increase in L. monocytogenes in VS remained below 3.4 log CFU/g throughout the storage. Furthermore, the equation μ_max-30S = {0.0059 × (T + 25.9983)}² was used to indirectly optimize shelf-life assessments for retail salads and provide reliable estimations (R² = 0.9951). The study successfully incorporated the back-propagation artificial neural network (BP-ANN) to validate the accuracy of traditional kinetic models in predicting the μ_max of L. monocytogenes. An R² greater than 0.97 was achieved across all training data. These findings offer invaluable insights for optimizing storage practices and improving food safety risk assessments for RTE salads.