Analysis of Steady-State Amperometric Biosensor with the Carbon Paste Electrode Using Physics-Informed Neural Network, AGM and TSM

The analytical model of the mediated bioelectrocatalysis is proposed in this paper. This model is based on a coupled steady-state nonlinear reaction–diffusion equations. In this model, we have also considered the effect of mediator leakage and the electrode surface fouling which majorly affects the working of the biosensor. We have utilized physics-based machine learning algorithm named as Physics-Informed Neural Network (PINN) for solving and training the nonlinear reaction–diffusion equation in CPE-based amperometric biosensor. The approximate analytical expressions for the nonlinear differential equations are obtained using Taylor’s Series Method and the Akbari–Ganji Method. These methods proved that they fit for all the parameters’ values appeared in the system in their specified range. With the steady-state biosensor current, the influence of the substrate concentration as well as the impact of the thickness of the enzyme layer on the biosensor response are analyzed. Analytical expressions for biosensor current, sensitivity and resistance are also presented and discussed briefly. The numerical solution for the nonlinear reaction–diffusion equation is obtained by MATLAB software, and it is noted that there is a satisfactory agreement when comparing the analytical solution with numerical solution for all possible parameter values. Furthermore, the influence of the diffusion and kinetic parameters on the biosensor current, sensitivity and resistance is analyzed.