UNCERTAINTY ASSESSMENT IN PREDICTIVE MICROBIOLOGY. PART 2: STUDY OF THE MODEL PARAMTER ESTIMATION UNCERTAINTY AS INDUCED BY EXPERIMENTAL ERRORS.
Food safety and quality are largely determined by the proliferation of pathogenic and spoilage micro-organisms during the life cycle of the product. In order to simulate and predict the microbial evolution in foods, mathematical models are developed in the field of predictive microbiology. At present, almost all models predicting the microbial load are deterministic: at a certain time instant the model predicts a single value for the microbial load. The uncertainty on the model outputs is not included. There is an urgent need for stochastic predictive microbiology, i.e., the construction of microbial evolution models that incorporate uncertainty at the level of measurements and parameters and predict the microbial load by a probability mass function. In this paper the influence of experimental errors on model parameter uncertainty is investigated by means of Monte Carlo analysis. As a case study, growth data of Escherichia coli K12 and Listeria monocytogenes lux at constant temperature are considered. The relationship between experimental uncertainty and parameter estimation uncertainty is established by repetition of the Monte Carlo analysis for a range of experimental uncertainties. In contrast to the non-linearity of the Baranyi growth model, results surprisingly show a linear relation between the experimental uncertainty and parameter estimation uncertainty. The linear relation is only valid for a range of realistic experimental uncertainties. The analogy of the results with common statistical methods is highlighted.
Poschet, F., Geeraerd, A.H., Scheerlinck, N., Bernaerts, K., Nicolaï, B.M. and Van Impe, J.F. (2001). UNCERTAINTY ASSESSMENT IN PREDICTIVE MICROBIOLOGY. PART 2: STUDY OF THE MODEL PARAMTER ESTIMATION UNCERTAINTY AS INDUCED BY EXPERIMENTAL ERRORS.. Acta Hortic. 566, 357-362
Monte Carlo, Baranyi and Roberts growth model, Escherichia coli K12, Listeria monocytogenes lux, stochastic modelling