Identifiability of the spatial SEIR-HCD model of COVID-19 propagation

TL;DR

This paper examines the identifiability of a spatial SEIR-HCD epidemic model for COVID-19, using sensitivity analysis and Bayesian methods to refine parameter estimation and highlight the need for additional data for diffusion coefficients.

Contribution

It introduces a combined Sobol sensitivity and Bayesian approach to improve parameter identifiability in a spatial epidemic model, emphasizing the necessity of supplementary information for certain parameters.

Findings

Parameter boundaries were reduced using sensitivity and Bayesian methods.

Diffusion coefficients require additional data for accurate identification.

The approach enhances understanding of spatial epidemic spread modeling.

Abstract

This paper investigates the identifiability of a spatial mathematical model of the spread of fast-moving epidemics based on the law of acting masses and diffusion processes. The research algorithm is based on global methods of Sobol sensitivity analysis and Bayesian approach, which together allowed to reduce the variation boundaries of unknown parameters for further solving the problem of parameter identification by measurements of the number of detected cases, critical and dead. It is shown that for identification of diffusion coefficients responsible for the rate of movement of individuals in space, it is necessary to use additional information about the process.