Neuro-Evolutionary Computing Paradigm for the SIR Model Based on Infection Spread and Treatment

Abstract

The intension of the present study is to solve the nonlinear biological susceptible, infected and recovered (SIR) models using Feed-Forward Artificial Neural Networks (FFANN) optimized with global search of genetic algorithm aided with rapid local search interior-point IP algorithms, i.e., FEANN-GAIP. An error-based cost function is formulated by exploiting FEANN models of differential equations and its associated conditions representing the SIR systems. The proposed FEANN-GAIP scheme is evaluated for three types of infection spread biological systems based on simple SIR, modified SIR and treatment-based SIR models. The reliability and correctness of the FEANN-GAIP scheme is substantiated with good agreement based on the results of the Adam numerical solver. The statistical outcomes further demonstrate the consistent precision, applicability, and robustness of the designed FEANN-GAIP based stochastic numerical solver.