Bayesian regularization neural network procedure for cancer cells in acquired immune deficiency syndrome model

Objective: This work presents the numerical solutions of the human immunodeficiency virus (HIV)-based dynamical infection mathematical model of cancer cells (DIMMCC), which is classified into cancer population cells, healthy cells, and HIV-infected cells. Based on the modeling of the HIV dynamics, growth of cancer cells, and immune response, DIMMCC can support scientists to recognize prospective therapeutic targets, predict the outcomes of treatment, and optimize mixture treatments, which finally enhance the capability to accomplish and treat cancer patients having HIV. Method: A Bayesian regularization neural network (BRNN) structure using the activation log-sigmoid function and twelve neurons is implemented to get the reliable and precise numerical performances of the HIV-based DIMMCC. A dataset is designed by the Runge–Kutta scheme, which reduces mean square error in training, testing and substantiations by 78%, 12%, and 10%. Results: The correctness of the designed procedure is observed by using the comparison of achieved and reference outputs, best training [Formula: see text] to [Formula: see text], and negligible absolute error found as [Formula: see text] to [Formula: see text]. To observe the efficiency and capability of the scheme, some test performances including state transition, regression, histogram, and correlation are presented. Novelty: The proposed BRNN, together with an activation log-sigmoid function and twelve neurons, has not been presented to solve the HIV-based DIMMCC.