Stepwise regression, Genetic Algorithm-Artificial Neural Network (GA-ANN) and Co-Active Neuro Fuzzy Inference System (CANFIS) were used to predict the effect of Satureja extracts (water and ethanol) on the population dynamics of Pseudomonas aeruginosa in a complex food system (Frankfurter sausage). The stepwise regression, GA-ANN and CANFIS were fed with four inputs: concentration (at five levels 0, 2000, 4000, 6000 and 8000 ppm), type of extract (water and ethanol), temperature (at three levels 5, 15 and 25°С) and time (1-20 days). The results showed that the stepwise regression was good for modeling the population dynamics of P. aeruginosa (R(2) = 0.92). It was found that ANN with one hidden layer comprising 14 neurons gave the best fitting with the experimental data, so that made it possible to predict with a high determination coefficient (R(2) = 0.98). Also, an excellent agreement between CANFIS predictions and experimental data was observed (R(2) = 0.96). In this research, GA-ANN was the best approach to simulate the population dynamics of P. aeruginosa. Furthermore, Satureja bachtiarica ethanol extract was able to reduce P. aeruginosa population, showing stronger effect at 5 °C and the concentration of 8000 ppm.
Keywords: Artificial neural network; Co-active neuro fuzzy inference system; Genetic algorithm; Modeling; Pseudomonas aeruginosa; Satureja bachtiarica.
Copyright © 2015 Elsevier Ltd. All rights reserved.