Dynamic Estimation of Epidemiological Parameters of COVID-19 Outbreak and Effects of Interventions on Its Spread

Hongzhe Zhang; Xiaohang Zhao; Kexin Yin; Yiren Yan; Wei Qian; Bintong Chen; Xiao Fang

To find appropriate priors, we note from Sun et al. 32 that the median incubation period of COVID-19 is estimated to be 4.5 days with interquartile range (IQR) 3.0-5.5 days, and the median delay between symptom onset and seeking care is 2 days with IQR 0-5 days in mainland China after January 18, 2020, the starting date of our analysis. Therefore the infectious period of COVID-19 ranges from 3 to 10.5 days. Accordingly, we set parameter µ to be uniformly distributed over ( 1 10.5 , 1 3 ). In addition, to the best of our knowledge, there is no literature on the duration from quarantine to removal for COVID-19 infected cases. Therefore, we collect data about 32 death cases and 22 cured cases in Wuhan from local newspapers. Details of these cases are given in Supplementary Table 2 . Among the death cases, the minimum duration of hospitalization is 1 day and the maximum is 40 days. The range of hospitalization for cured cases is from 6 to 30 days. For COVID-19 infected cases in Wuhan, the percentage ratios of death and cure are 5.8% and 94.2%, respectively 33 . Accordingly, we roughly estimate the duration from quarantine to removal in the SIQR model to have range from 5.7 to 30.6 days using weighted averages, and we set parameter γ to be uniformly distributed over ( 1 30.6 , 1 5.7 ). Non-informative flat priors are adopted for the rest parameters so that (10) in the following summarizes our empirical prior settings.