A radial basis Bayesian regularization neural network process for the malaria disease model

Purpose: Malaria is one of the dangerous infectious disease produced through Plasmodium parasites, which is transmitted by the bite of a diseased Anopheles mosquito. The aim of this study is to solve the malaria disease model by using one of the stochastic computing neural network structures. The nonlinear system is presented into the populations of the host and vector using the insecticides and treatment. Method: A radial basis neural network uses the radial basis in the hidden layer. The performance of optimization is judged via Bayesian regularization for presenting the solutions of the model. The construction of the data is performed through the explicit Runge-Kutta that decreases the mean square error by adjusting the data for authentication 0.12, testing 0.15, and training 0.73. Results: The accurateness of designed stochastic technique is observed based on the matching of the obtained and the published solutions. Moreover, the negligible absolute error performances 10⁻⁰⁵ to 10⁻⁰⁷, and best training values 10⁻⁰⁹ to 10⁻¹² also support the exactness of the solver. In addition, the capability of the designed scheme is validated through different values based state transition, regression, and error histogram. Novelty: The designed stochastic computational radial basis neural network procedure along with the optimization of Bayesian regularization has never been applied before to solve the malaria disease system.