Efficient modeling approach for simulating multi-physics responses of an ion-sensitive field-effect transistor using artificial neural networks

The application of neural networks has been investigated in many fields, but the feasibility of estimating the nonlinear multi-physics behavior in a real-time environment by applying neural networks is not sufficiently matured. This study presents an efficient modeling approach based on an artificial neural network (ANN) to precisely predict the multi-physics response of an ion-sensitive field-effect transistor (ISFET), and it is shown to be a promising technique for measuring the ion concentration in solution. An ANN-based ISFET (NN-ISFET) model is designed in this study to represent the nonlinear relationship between drain current and input parameters. A machine learning technique along with a new feasible algorithm is introduced to optimize the ISFET model, and the strong approximation ability of the NN-ISFET model enables the rapid and accurate prediction of the DC characteristic curves of the transistors. The experimental analysis validates the effectiveness of the proposed methodology and reveals that the proposed modeling approach can save approximately 98% of the computational cost as compared with conventional commercial software. Additionally, the CPU time achieved using the proposed model is less than 0.1 s, unlike commercial software with high time and memory consumption.