Data mining and fuzzy modelling of high pressure inactivation pathways of Lactococcus lactis

A mathematical model to predict lethal and sublethal pressure effects on Lactococcus lactic was established through combined use of Principal Component Analysis (PCA) and Fuzzy Logic. The model was based on data comprising pressure inactivation kinetics with 64 combinations of the parameters pressure, temperature, pH, and buffer composition, samples were analysed with respect to 5 physiological states describing lethal or sublethal injury of L. lactis, i.e. viable cell counts (CFU), undamaged cell counts (CFUsub), membrane integrity (MI), metabolic activity (MA) and the activity of the membrane bound enzyme LmrP (LmrP).Correlations found by PCA were used to generate a bi-layer fuzzy model using clustering methods and rule oriented statistical analysis as well as the physiological states CFU and LmrP as autonomous output variables. The result of these variables is used to accurately predict dependent output variables MA, CFUsub, and MI taking into account the combined effects of the inactivation process.Industrial relevanceThe study provides a predictive model for the inactivation of an industrially relevant micro-organism. Predictive models are useful for process design on the industrial scale. Further, our model provides information about intermediate steps of high pressure inactivation.