A numerical treatment through Bayesian regularization neural network for the chickenpox disease model.

Area of Science:

• Epidemiology
• Computational Biology
• Artificial Intelligence

Background:

• Chickenpox (varicella) is a contagious disease requiring accurate mathematical modeling for control.
• Understanding disease dynamics involves compartmental models categorizing populations (susceptible, vaccinated, infected, etc.).

Purpose of the Study:

• To develop and apply a novel artificial neural network (ANN) framework for the numerical solution of a chickenpox disease model.
• To assess the efficacy of Bayesian regularization within an ANN for solving complex epidemiological models.

Main Methods:

• A single hidden layer artificial neural network with Bayesian regularization was constructed.
• The dataset was generated using the Runge-Kutta technique, with data split for training (76%), validation (12%), and testing (12%).
• A logistic sigmoid fitness function and thirty neurons were employed in the ANN architecture.

Main Results:

• The ANN model demonstrated high accuracy, with negligible absolute errors ranging from 10^-04 to 10^-06.
• Excellent performance was achieved, indicated by mean square error values between 10^-09 and 10^-11.
• Model reliability was confirmed through result matching, regression analysis, and error histograms.

Conclusions:

• The proposed artificial neural network framework with Bayesian regularization offers a reliable and accurate method for solving the chickenpox disease model.
• This represents the first application of this specific ANN architecture and optimization technique to the chickenpox epidemiological model.