COMBINING DATA ASSIMILATION AND THE SIR INFECTION EQUATION FOR BETTER UNDERSTANDING OF COVID-19

Abstract

The outbreak of the new coronavirus pandemic (COVID-19) has created a global health crisis, and a significant challenge to mitigate its spread and impact on global communities. The SIR equation has been widely used as a theoretical model to simulate the course of an infection, however, its parameters cannot be known precisely due to the time lag between infection and onset and infectivity in the not-onset state. In this document, we propose a solution to the SIR infection equation using data assimilation to tackle the issue of parameter estimation in the present case. We integrate the SIR equation and observation equation within the framework of data assimilation, and estimate the unknown parameters using the maximum likelihood method. Data assimilation is considered effective in estimating the true value of unknown parameters through Bayesian inference. The state-space model is used to improve the accuracy of parameter estimation, generating virtual numerical simulations under different conditions and comparing the observation data on newly infected individuals, dead individuals, and severely ill individuals. Our study evaluates the effectiveness of data assimilation in obtaining accurate estimates of the infection rate, recovery rate, initial number of infected people, observation coefficient, and standard deviation of observation data noise