A machine learning computational approach for the mathematical anthrax disease system in animals.

Area of Science:

• Veterinary epidemiology
• Computational biology
• Machine learning applications in disease modeling

Background:

• Anthrax poses a significant threat to animal health.
• Mathematical models are crucial for understanding disease dynamics.
• Numerical solutions are needed for complex disease systems.

Purpose of the Study:

• To develop and present a novel machine learning stochastic procedure for numerical solutions of the animal anthrax disease system.
• To model the disease dynamics including susceptible, infected, recovered, and vaccinated states.

Main Methods:

• A Runge-Kutta solver was used for dataset generation, with data split into 78% training, 12% testing, and 10% verification.
• The stochastic computing technique employed a logistic sigmoid activation function, a single hidden layer with 27 neurons, and Bayesian regularization for optimization.

Main Results:

• The proposed method achieved high accuracy, with absolute errors ranging from 10^-5 to 10^-8.
• Excellent training performance was observed, with errors as low as 10^-10 to 10^-12.
• Statistical metrics including regression coefficient and error histogram confirmed the reliability of the approach.

Conclusions:

• The developed machine learning stochastic procedure provides accurate numerical solutions for the animal anthrax disease system.
• This novel approach enhances the reliability of disease modeling and analysis.
• The method's effectiveness is validated by its high accuracy and statistical performance.