Validation of a polynomial regression model: the thermal inactivation of Bacillus subtilis spores in milk

Abstract

Aims: The predicted survival of Bacillus subtilis 168 spores from a polynomial regression equation was validated in milk. Methods and Results: Bias factor suggested as an index of model performance was used to validate the polynomial model predictions in ultrahigh temperature (UHT) treated and sterilized whole and skim milk. Model predictions were fail safe, predicting higher D-values (decimal reduction times) in buffer than actually noted in milk. Conclusions: The D-values for spores were lower in milk as compared with those predicted in potassium phosphate buffer contrary to the popular expectation of better spore survival in complex food systems. The Bias factor, a quantitative measure of the model performance, indicated that on average the model predictions exceed the observations by 40% in the case of whole milk and by 60% in the case of skim milk. Significance and Impact of the Study: The present work is an attempt to ascertain the extent of reliability that one can safely place in polynomial model predictions, without compromising on the safety or palatability of foods where it is eventually intended to be applied. The work has also highlighted the differences in the thermal inactivation pattern of spores in buffer and in milk with a possible influence of the various constituents of milk. The work will assist the dairy industry to better design thermal processes to ensure longer shelf life of dairy foods.

Modelling the effect of water immersion thermal processing on polyacetylene levels and instrumental colour of carrot disks

“…A predictive model may only be reliably used in decision making when tested and validated (Jagannath & Tsuchido, 2003). Validation is a vital step that reveals the applicable range of a model and the limits of its performance (Gabriel, 2008).…”

Application of principal component and hierarchical cluster analysis to classify different spices based on in vitro antioxidant activity and individual polyphenolic antioxidant compounds

“…The mathematical predictive model assessments were carried out (Jagannath & Tsuchido, 2003) by calculating the model performance indices, accuracy factor (AF), bias facto (BF) (Patras, Brunton, r Tiwari, & Butler, 2009;Ross, 1996):…”

Responses of E. coli O157:H7, L. monocytogenes 1/2 c and Salmonella enteritidis to pH, aw and temperature stress combinations

“…The IFR (2009) models were established using pathogen strains different from those used in this study. In fact, variations were observed in predicted and observed values even when the same reference organisms were used to validate the efficacies of predictive models (Carrasco et al, 2006;Jagannath & Tsuchido, 2003).…”