Modelling of the kinetics of red tilapia (Oreochromis spp.) viscera enzymatic hydrolysis using mathematical and neural network models

Abstract

The present work modelled the enzymatic hydrolysis of red tilapia (Oreochromis spp.) viscera with Alcalase® 2.4 L in both 0.5 and 5 L reactors. The best conditions for the enzymatic hydrolysis were 60°C and pH 10. The product inhibited the enzymatic hydrolysis, and the enzyme deactivated following second-order reaction. K_M and K_p from a secondary plot of K_M^app as a function of inhibitor concentration, and k_2, p, and k_3 were found by non-linear regression. While the obtained parameters modelled the 0.5 L reactor well, it did not model the 5 L reactor, probably because of unconsidered fluid dynamics in the model. To have a better modelling, a neural network (tensorflow.keras.models module) was built and trained. The neural network modelled the enzymatic hydrolysis of red tilapia at several concentrations of substrate and enzyme. This result proved that neural networks are a powerful tool for modelling biological processes.