Modeling and Simulation of Enzymatic Biosensor for Detecting Aflatoxin B1 Using Artificial Neural Network

Aflatoxin B1 (AFB1) is one of the most toxic Aflatoxins that contaminates agricultural products and causes deathlike effects on human health. Determination of AFB1 in food by biosensors is fast, low cost and accurate. In this paper, modeling and simulation of chemical reactions in the AFB1 potentiometric biosensor is performed to determine the optimal reaction rate constants. Enzymatic reactions are simulated using COMSOL software and reaction rates are optimized by Artificial Neural Network (ANN) and Genetic Algorithm (GA). The fitness function of GA is defined by deploying ANN. The data generated during the simulation step were used to train and evaluate the performance of the neural network. Compared with experimental data, COMSOL model simulated biosensor response with MAPE equal to 0.1023 %. In addition trained ANN with 5-48-1 structure predicted biosensor response with MAPEs equal to 0.7074 %, 0.9458 %, 0.7473 % and 0.7492 % for train, validation, test and total data sets respectively. Reaction rates were optimized by Artificial Neural Network (ANN) and Genetic Algorithm. Modeling results showed that trained Neural Network using Genetic Algorithm optimized reaction rates has the lowest MAPE equal to 0.0026 % compared with other models in prediction of AChE enzyme inhibition by AFB1.