An Improved SEIR Model for Reconstructing the Dynamic Transmission of COVID-19

ABSTRACT

With the recent outbreak of coronavirus disease 2019 (COVID-19), human life and the world economy have been severely affected, the propagation and scale of COVID-19 is top of mind for everyone. To reconstruct the development trend of COVID-19, we investigate the issue of the epidemic spreading process under vigorous non-pharmaceutical interventions. Here, an improved Susceptible-Exposed-Infectious-Recovered (SEIR) model with dynamic variables (i.e., health exposure individuals and close contacts) is proposed to predict the scale of COVID-19 and its dynamic evolution. We assume that the number of contacts and the reproduction number of COVID-19 changes dynamically over time. Then a gradient descent method is applied to estimate the effective reproduction number. We use the proposed model to reconstruct the dynamic transmission of COVID-19 in Chongqing between 14 January and 24 March 2020. The results show a similar development trend with a real-world epidemic. Our work has important implications when considering strategies for continuing surveillance and interventions to eventually contain outbreaks of COVID-19.