Purpose: The impact of both, variability and uncertainty was quantified in the microbiological shelf-life estimation of food products.
Methods: Various growth kinetics, within specific food items, were modeled, using the same environmental conditions of temperature, pH and water activity. The study was thus run with Listeria monocytogenes as bacterial model. Three products were studied: pâté, cured herring and smoked chicken. The numbers of growth kinetics (n=3; 10), dates (n=5; 10) and points (n=1; 3) were considered to evaluate the variability and uncertainty related to modeling. The estimation of growth parameters were characterized with a Non-Linear Mixed Effect models, based on the stochastic version of Expectation-Maximisation Algorithm (SAEM). These estimates were obtained with the Monolix software, and were then used for simulation with the 2D Monte Carlo calculation in the Sym’Previus decision making tool (www.symprevius.org).
Results: The proposed stochastic approach calculates the probability to overpass with precision a critical value during product shelf-life. While the demonstration was done with L. monocytogenesspecies, the same calculations can be applied to predict the quality indicators and spoilage bacteria behavior.
Significance: Variability and uncertainty in the initial contaminations and in the chemical parameters, such as pH and water activity, could be integrated in the developed stochastic model to determine the probability to exceed microbial criteria during food storage. The approach is particularly useful to estimate food product shelf-lives, to run QRA, to determine and target the significant microbial and chemical quality controls in HACCP.