Statistical versus artificial intelligence -based modeling for the optimization of antifungal activity against Fusarium oxysporum using Streptomyces sp. strain TN71

A Streptomyces sp. strain TN71 was isolated from Tunisian Saharan soil and selected for its antimicrobial activity against phytopathogenic fungi. In an attempt to increase its anti-Fusarium oxysporum activity, GYM + S (glucose, yeast extract, malt extract and starch) culture medium was selected out of five different production media. Plackett-Burman design (PBD) was used to select yeast extract, malt extract and calcium carbonate (CaCO3) as parameters having significant effects on antifungal activity, and a Box-Behnken design was applied for further optimization. The analysis revealed that the optimum concentrations for the anti-F. oxysporum activity of the tested variables were yeast extract 5.03 g/L, malt extract 8.05 g/L and CaCO3 4.51 g/L. Artificial Neural Networks (ANNs): the Multilayer perceptron (MLP) and the Radial basis function (RBF) were created to predict the anti-F. oxysporum activity. The comparison between experimental and predicted outputs from ANN and Response Surface Methodology (RSM) were studied. The ANN model presents an improvement of 14.73%. To our knowledge, this is the first work reporting the statistical versus artificial intelligence -based modeling for the optimization of bioactive molecules against mycotoxigenic and phytopathogenic fungi.