Numerical Investigations through ANNs for Solving COVID-19 Model

Area of Science:

• Computational epidemiology
• Mathematical modeling of infectious diseases
• Artificial intelligence in public health

Background:

• Understanding COVID-19 transmission dynamics is crucial for effective public health interventions.
• Traditional epidemiological models often require significant computational resources or simplifying assumptions.
• Developing efficient and accurate numerical methods for epidemic modeling remains an active research area.

Purpose of the Study:

• To investigate the application of artificial neuron networks trained with Levenberg-Marquardt backpropagation (ANNs-LMB) for modeling COVID-19 spread.
• To assess the accuracy and performance of the ANNs-LMB approach in approximating numerical solutions for the COVID-19 spreading model.
• To compare the results obtained from ANNs-LMB with a reference dataset generated by the Runge-Kutta scheme.

Main Methods:

• Implementation of artificial neuron networks (ANNs) utilizing the Levenberg-Marquardt backpropagation (LMB) training algorithm.
• Data partitioning into training (80%), validation (10%), and testing (10%) sets for robust model evaluation.
• Numerical solution approximation using ANNs-LMB and comparison with reference solutions derived from the Runge-Kutta method.

Main Results:

• The ANNs-LMB scheme successfully generated approximate numerical solutions for the COVID-19 spreading model.
• Performance evaluation demonstrated the viability of ANNs-LMB by minimizing mean square error (M.S.E).
• Error histograms, regression, and correlation analyses confirmed the reliability and accuracy of the model's solution dynamics.

Conclusions:

• Artificial neuron networks trained with Levenberg-Marquardt backpropagation offer a viable and accurate computational approach for modeling COVID-19 transmission.
• The ANNs-LMB method provides a robust alternative for approximating epidemiological model solutions, showing good agreement with established numerical schemes.
• This study highlights the potential of AI-driven models in enhancing our understanding and prediction of infectious disease dynamics.