Hydrolysis of lactose: estimation of kinetic parameters using Artificial Neural Networks

Abstract

The analysis of any kinetic process involves the development of a mathematical model with predictive purposes. Generally, those models have characteristic parameters that should be estimated experimentally. A typical example is Michaelis-Menten model for enzymatic hydrolysis. Even though conventional kinetic models are very useful, they are only valid under certain experimental conditions. Besides, frequently large standard errors of estimated parameters are found due to the error of experimental determinations and/or insufficient number of assays. In this work, we developed an artificial neural network (ANN) to predict the performance of enzyme reactors at various operational conditions. The net was trained with experimental data obtained under different hydrolysis conditions of lactose solutions or cheese whey and different initial concentrations of enzymes or substrates. In all the experiments, commercial beta-galactosidase either free or immobilized in a chitosan support was used. The neural network developed in this study had an average absolute relative error of less than 5% even using few experimental data, which suggests that this tool provides an accurate prediction method for lactose hydrolysis