Neural networks provide superior description of Giardia lamblia inactivation by free chlorine

Use of conventional models to describe data on microbial inactivation during disinfection suffers from limitations with respect to flexibility and direct quantitative incorporation of water quality variables. This paper develops an approach to analysis of such data using neural networks (NNs). Using the data on free chlorine inactivation of Giardia lamblia previously reported, it was found that the use of an NN with a single hidden layer and four hidden neurons provided a superior (better) fit to the data with a reduced number of model parameters when compared to the fitting of this data using a conventional approach. Therefore, the use of NN models should be considered in the future assessment of microbial inactivation during disinfection. Incorporation of additional facets of the disinfection process, such a disinfectant decay, needs to be considered in subsequent development of this approach.