Prediction of the COVID-19 outbreak based on a realistic stochastic model

Yuan Zhang; Chong You; Zhenghao Cai; Jiarui Sun; Wenjie Hu; Xiao-Hua Zhou

The containment time of the outbreak is defined as the moment when the number of active virus carriers is first less than a threshold T c . Noting that the spread of the disease is due to the existing active virus carriers in the proposed model, thus when the number of the active virus carriers decreases to a certain level, the new infections are limited, which means the epidemic is under control. Figure 5 shows the 95% confidence interval of the containment time of the outbreak for each region when T c = 10. Among the six regions in this study, Shanghai is predicted to have the earliest containment time of February 21, while the containment time in Guangdong is predicted to be the latest, around March 7. From the daily report of NHC of China, new infections in Shanghai from February 17 to February 21 are 0,0,0,1,0; and new infections in Guangdong in the same period are 6,3,1,1,6, which potentially supports our prediction. Notice that in sensitivity analysis on γ A , we find that when the recover rate of asymptomatic patients γ A gets smaller, the expected containment time is longer. This is because smaller γ A means that infected asymptomatic patients need 17 . CC-BY-NC-ND 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.