Application of Artificial Neural Networks to Predict Inhibition in Probiotic Experiments

Artificial neural networks (ANNs) provide a modeling approach that can be used in the vitro stages of probiotic studies. The aim study was to evaluate ability multilayer perceptron (MLP) and radial-basis function (RBF) ANNs predict inhibition level indicator bacteria co-culture experiments performed at various initial concentrations. In both types networks, time, concentrations L. lactis Aeromonas spp. were input variables concentration output value. construction models, different numbers neurons hidden layer, activation functions examined. performance developed MLP RBF models tested with root mean square error (RMSE), coefficient determination (R2) relative (e) statistical analysis. Both ANN showed strong agreement between predicted experimental values. However, higher accuracy efficiency than models. results indicated this successfully co-cultured determine bacterial design further experiments.