Estimation of the Unreported Infections of COVID-19 based on an Extended Stochastic Susceptible-Exposed-Infective-Recovered Model

Zhu, L.; Dong, W.; Sun, Q.; Vargas, E. A. H.; Du, X.

In this paper, an innovative SEIR(Susceptible-Exposed-Infective-Recovered) model is proposed to estimate the true infectivity and lethality of the COVID-19 epidemic in Wuhan, China. Segmented parameters are used in the model to prove the effectiveness of improved public health interventions such as city lockdown and extreme social distancing.And the generally polynomial chaos method is used to increase the reliability of the model results in the case of parameter estimation. The accuracy and validity of the proposed SEIR model are proved according to the official reported data.Also, according to the epidemic trend reflected by the model, the effectiveness and timeliness of the epidemic prevention policies formulated by the government can be reflected. © 2021 IEEE.