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

There is a total of 9 model parameters in the proposed model for each selected region. They are λ IN , θ E , ρ, q, γ IH , γ A , r s , r q and r H , among which, r s , r q and r H are related to the clinical characteristics of the disease and can be prefixed through existing studies. To be more specific, r H is the inverse of the average time from symptoms onset to diagnosis, r s is the inverse of the mean incubation period, while r q is the inverse of mean difference between infectious period and serial interval. Based on preliminary trials, we find there is very limited information of γ A which can be obtained from the data, and the estimate is highly influenced by the choices of prior. A possible explanation towards it is that γ A is less related to the observations. Hence, instead of estimating γ A with large uncertainty, we prefix γ A = 1/10. Sensitivity analysis is conducted on the different choices of γ A .