新型冠状病毒肺炎早期时空传播特征分析

通过最新公布的流行病学数据估计了易感者-感染者模型参数,结合百度迁徙数据和公开新闻报道,刻画了疫情前期武汉市人口流动特征,并代入提出的支持人口流动特征的时域差分方程模型进行动力学模拟,得到一些推论:1)未受干预时传染率在一般环境下以95%的置信度位于区间0.2068,0.2073],拟合优度达到0.999;对应地,基本传染数R0位于区间2.5510,2.6555];极限环境个案推演的传染率极值为0.2862,相应的R0极值为3.1465;2)百度迁徙规模指数与铁路发送旅客人数的Pearson相关系数达到0.9108,有理由作为人口流动的有效估计;3)提出的模型可有效推演疫情蔓延至外省乃至全国的日期,其中41.38%的预测误差≤1 d,79.31%的预测误差≤3 d,96.55%预测误差≤5 d,总体平均误差约为2.14 d.

Analysis on early spatiotemporal transmission characteristics of COVID-19

In this paper,a simple susceptible-infected(SI)model is build for simulating the early phase of COVID-19 transmission process.By using the data collected from the newest epidemiological investigation,the parameters of SI model is estimated and compared with those from some other studies.The population migration data during Spring festival in China are collected from Baidu.com and also extracted from different news sources,the migration characteristic of Wuhan city in the early phase of the epidemic situation is captured,and substituted into a simple difference equation model which is modified from the SI model for supporting migrations.Then several simulations are performed for the spatiotemporal transmission process of COVID-19 in China.Some conclusions are drawn from simulations and experiments below.1)With 95%confidence,the infection rate of COVID-19 is estimated to be in a range of 0.2068–0.2073 in general situation,and the corresponding basic reproduction number R0 is estimated to be in a range of2.5510–2.6555.A case study shows that under an extreme condition,the infection rate and R0 are estimated to be 0.2862 and 3.1465,respectively.2)The Pearson correlation coefficient between Baidu migration index and the number of travelers sent by railway is 0.9108,which indicates a strong linear correlation between them,thus it can be deduced that Baidu migration index is an efficient tool for estimating the migration situation.3)The epidemic arrival times for different provinces in China are estimated via simulations,specifically,no more than 1 day within an estimation error of 41.38%;no more than 3 days within an error of 79.31%,and no more than 5 days with an error of 95.55%.An average estimation error is 2.14 days.