Artificial neural network and response surface methodology for modeling and optimization of activation of lactoperoxidase system

Abstract

In the present study, the multi-component lactoperoxidase system (LPS) is used for improving milk safety and requires thiocyanate (SCN−) as a substrate for the generation of antimicrobial hypothiocyanite (OSCN−). The influence of four independent variables for activation of lactoperoxidase system on the improving the quality of raw goat milk were investigated and optimized using an artificial neural network and response surface methodology on the growth of total coliform count and bacterial count. The two models' predictive capabilities were compared in terms of root mean square error, mean absolute error, standard error of prediction, absolute average deviation, and coefficient of determination based on the validation data set. The results showed that properly trained artificial neural network model is more accurate in prediction than the RSM model. The optimum conditions were a temperature of 25 °C, storage time of 10 hr, NaSCN of 30 ppm and hydrogen peroxide of 18 ppm. For these conditions, an experimental total coliform count of 4.51 × 102cfu/mL and total bacteria count of 5.44× 104cfu/mLwas obtained, which was in reasonable agreement with the predicted content.The results indicate that the model is in substantial agreement with current research, and activating the LP System can extend the storage period of goat milkfor up to10hr when stored at 25 °C.The results revealed no significant differences in milk composition (protein content, fat content, lactose content, total solids, moisture content and ash content) were observed among activated and control goat milk samples.

Keywords

• Goat milk
• Lactoperoxidase system
• Total coliform count
• Total bacterial count
• Response surface methodology (rsm)
• artificial neural network (ANN)