Spatially Explicit Modeling of 2019-nCoV Epidemic Trend based on Mobile Phone Data in Mainland China

Xiaolin Zhu; Aiyin Zhang; Shuai Xu; Pengfei Jia; Xiaoyue Tan; Jiaqi Tian; Tao Wei; Zhenxian Quan; Jiali Yu

For parameters 1 , 2 , and a, we used the daily cumulative confirmed cases up to February 10, 2020 to retrieve their optimal values to reflect the current dynamics in each city. We first estimated the optimal value of 2 and a of Wuhan since its epidemic model only has one transmission rate 2 . Then, the estimated 2 was used as a prior parameter for the estimation of 1 and a for individual prefecture-level cities. The Nelder-Mead algorithm 17 was employed to estimate parameters through minimizing the sum of squared differences between the simulated and actual daily cumulative confirmed cases. To better capture the current trend, the number of daily cumulative cases were also used to weight the samples in parameter retrieval. Since the epidemic model is highly sensitive to the initial infectious number I0, and the reported initial infectious number often has large uncertainty, we treated the initial infectious number 0 as another parameter to be estimated together with 1 and 2 . Specifically, we used daily cumulative confirmed cases of each city from January 25 to February 10, 2020 to retrieve the parameters and assumed January 20, 2020 as the start point when massive inter-city mobility happened before the Spring Festival. The goodness of model fitting was assessed by comparing the number from model simulation and reported cases. The trained model was further validated using the reported data on February 11, 2020.