Comparison of Arrhenius model and artificial neuronal network for predicting quality changes of frozen tilapia (Oreochromis niloticus)

•Ice crystal formation combined with protein denaturation to predict shelf life was first reported.•The prediction accuracy of the kinetic model was decreased at later stages of storage.•IFIλmax could be accurately described the quality changes of tilapia.•ANN could successfully predict the quality changes throughout the whole period.•It is the first time to use K value to predict the freshness range of other indexes.