Abstract:

In order to compare the goodness-of-fit of growth models for the total bacterial count in vacuum-packaged
chicken breakfast sausage, the primary models Baranyi, modified Gompertz and modified Logistic, and the secondary
models Arrhenius and square root equations were chosen for conducting following studies, respectively. They were
compared by calculating their indices (such as root mean square error (RMSE), R2, Akaike information criterion and
Bayesian information criterion), and the characteristic values and residuals of the shelf life model. Results showed that the
Baranyi model exhibited the best performance indices and the highest maximum specific growth rate (μmax) and provided
smaller shelf life residuals, while the modified Gompertz model was better in optimizing the secondary model and the
modified Logistic model presented the wrong characteristic values. Thus the goodness-of-fit of the Baranyi model was the
best, followed by the modified Gompertz model, and the modified Logistic model showed the worst goodness-of-fit. The
comparison of Arrhenius equation and square root equation did not draw a conclusion.

Key words: prediction model, goodness-of-fit, temperature, comparison