Neuro-evolutionary computing paradigm for two strain COVID-19 model

Abstract

The hybrid computational approaches have captured the interest of researchers over the past few decades. This paper examines the evolutionary integrated methodology, i.e. IGASQP of artificial neural networks (ANNs), to solve the nonlinear two-strain COVID-19 pandemic model (TSCPM). This approach exhibits the integrated characteristics of a global algorithm, e.g. the genetic algorithm with local search sequential quadratic programming (IGA-SQP). COVID-19 is a respiratory disease prompted by a mutable ribonucleic acid (RNA) virus. Around the world, variants with many attributes that might influence transmissibility emerged in December 2020. To address this new variant disease, a five-class mathematical model is constructed that comprises susceptible, vaccinated, strain-1 infected, strain-2 infected, and recovered humans based on the disease status of each individual. A mean squared error-based objective (fitness) function is established using TSCPM in terms of ANNs. The IGA-SQP results are provided and compared with the solution of the Adams numerical approach (ANA), and the corresponding absolute error is calculated. The performance indicators are established for many simulations runs that certify the flexibility, robustness, and effectiveness of IGA-SQP to solve TSCPM.