An Intelligent Framework for Modeling Nonlinear Irreversible Biochemical Reactions Using Artificial Neural Networks

Abstract The objective of this paper is to perform a comparative performance analysis of three different algorithms, namely Levenberg-Marquardt (LM), Bayesian Regularization (BR), and Scaled Conjugate Gradient (SCG), for nonlinear irreversible biochemical reaction (NIBR). These algorithms are based on artificial neural network (ANN) techniques. Nonlinear irreversible biochemical reactions are described by a set of nonlinear ordinary differential equations (ODE’s) that take into account the concentration of different compounds, such as products , substrate, enzyme-substrate complex, and enzyme-product complex. To develop the Backpropagation Levenberg Marquardt artificial neural networks (BLM-ANN) model, a dataset is created using the RK4 method, incorporating diverse NIBR models with varying reaction rates. A comparative analysis is presented, focusing on the performance of algorithms in handling the trained dataset. The design of the ODE’s system of the proposed model is done using the MATLAB/Simulink environment. To analyze the developed model, the artificial neural network toolbox is employed. The reliability and efficiency of the BLM-ANN model were evaluated by training, validation, and testing phase, it demonstrated an accurate prediction of irreversible biochemical reaction when compared to approximate result. The effectiveness of the proposed BLM-ANN model is measured using various metrics, such as state transition (ST), error-histograms (EH), mean square error (MSE), and regression analysis (R). These evaluation showed the value of the numerical solution provided by the BLM-ANN model for irreversible biological reactions. A comparison between the proposed BLM-ANN model and reference solution for the NIBR model demonstrated its effectiveness with observed convergence up to 10 ⁻⁸ to 10⁻³ .