ABSTRACT
As the novel coronavirus disease 2019 (COVID-19) has become a Public Health Emergency of International Concern, it is greatly significant to accurately predict the disease's incoming trend. Herein, we performed a stage-rolling Susceptible-Exposed-Infectious-Recovered (SEIR) model to measure the evolution of the basic reproduction number of COVID-19, based on the number of confirmed infections announced by the National Health Commission of the People's Republic of China. We assumed that the infected number under the spreading of infectious diseases will generally experience two different stages. In the first stage, due to the public's ignorance of the severity and harmfulness of the disease, the infected population grows exponentially and the process of disease transmission can be considered as the classic SEIR model with a constant basic reproduction number. Consequently, with limited awareness of the epidemic situation, there is a lack of effective preventive measures to control disease transmission. In the second stage, various control measures and medical resources are introduced in succession by the government, as well as the public gradually takes effective preventions (e.g. keep social distance and wear masks) based on the knowledge of the disease transmission. Collectively, the infected population grows much slower than the first stage. We performed a stage-rolling SEIR model, in which the basic reproduction number changes every day. Based on this model, the number of daily basic reproduction is estimated from the daily new infection number. We found that the daily basic reproduction number is expected to decline continually until it is less than 1, which means the eradication of the disease. Leveraging the evolution of the basic reproduction number, we extrapolate the incoming daily basic reproduction number, based on which we further predict the incoming trend of COVID-19 spreading in terms of the daily infection number. Our predictive model estimates that at the end of the epidemic, the total number of infections in China is nearly 14000 except for Hubei Province, and 32000 except for Wuhan city. We also found that in most parts of China, the number of newly confirmed infections increases linearly rather than exponentially before the day of "Wuhan travel restrictions", implying that the prevention and containing the infected people from Wuhan at the eve of the Spring Festival has been effective from the beginning.