A New Probabilistic Approach in Predictive Microbiology

Assessment of growth and tenacity of bacteria during food production processes and storage is vital for food security. Predictive microbiology forecasts population dynamics of microorganisms in food or feed, using mathematical models. It is a basic component of microbial risk assessment, and it can help to prevent foodborne disease outbreaks, because it assists by detecting probable contamination sites, and defining monitoring points. Predictive models that do not include variability of the modeled system are called deterministic. They only yield point estimates, and cannot reproduce the full range of possible kinetics. Today, most models in predictive microbiology are deterministic. Substitution of deterministic parameters with probability distributions yields probabilistic models. The use of probabilistic models in predictive microbiology supports a more profound risk assessment of the food or feed supply chain than deterministic models.

The goal of existing models in predictive microbiology is to gain understanding of population dynamics. The global error is minimized by fitting the parameters with pooled experimental data. For a risk assessment it is essential to predict the bacterial count at a single timepoint (at the end of the process chain) as precise as possible, i.e., the local error has to be minimized. The New Probabilistic Model in Predictive Microbiology (NPMPM) is based on a new approach for including variability and uncertainty into modeling of microbial growth. The NPMPM was developed for risk assessment of bacterial contamination in the food supply chain. We will introduce the NPMPM by the example of a contamination of the milk supply chain with Listeria monocytogenes.