Can neural networks estimate parameters in epidemiology models using real observed data?

Abstract

The primary objective of this study is to address the challenges associated with estimating parameters in mathematical epidemiology models, which are crucial for understanding the dynamics of infectious diseases within a population. The scope of this research includes the development and application of a two-phase neural network for parameter estimation, specifically within the context of epidemic compartmental models. This study presents a novel approach by integrating an extreme learning machine with a heuristic population-based optimization method within a two-phase neural network framework. The networks are driven by a heuristic population-based optimization method, enhancing the accuracy and efficiency of parameter estimation in mathematical epidemiology models. The effectiveness of the method is validated using actual COVID-19 data provided by the Turkish Ministry of Health. The data includes cases categorized as Susceptible, Exposed, Infected, Removed, and Deceased, which are crucial components of epidemic compartmental models. The obtained results highlight the capability of the proposed method to provide insights into the spread of infectious diseases by offering reliable estimates of model parameters. This, in turn, supports better understanding and forecasting of disease dynamics. The methodology provides a significant contribution to the field by offering a new, efficient technique for parameter estimation in epidemiological models.