基于引入隐形传播者的SEIR模型的COVID-19疫情分析和预测

在传统SEIR模型基础上,引入了“隐形传播者”的概念,并利用2020年1月25日至2月22日的COVID-19疫情数据进行模型拟合,并对结果进行分析。同时,利用拟合好的模型对2020年2月22日之后的演化情况进行仿真。结果显示,引入隐形传播者的SEIR模型在拟合和预测性能上有显著提升,降低了50%～70%的拟合误差;拟合系数表明,在疫情前期和后期携带病毒人群占潜伏者的比例分别为30%和5%,被确诊的概率由7%上升为40%,核酸检测技术趋于成熟;隐形传播者初期人数约为70 000,在国家有效管控下,目前控制在2 500名附近;预测3月中旬为“拐点”,4月底居民可恢复正常生活,最终累计确诊数在100 000左右。

Assessment and Prediction of COVID-19 Based on SEIR Model with Undiscovered People

In this paper the concept of ‘undiscovered people’ is introduced into traditional SEIR model for the research of COVID-19. The model fitting with result analysis is performed by utilizing COVID-19 data from 25 January 2020 to 22 February 2020. And then the fitted model is used to simulate the evolution after February 22, 2020. The results show that the introduction of ‘undiscovered people’ has a significant improvement in fitting and prediction performance, reducing the fitting error by 50% to 70%. The fitting coefficient shows that the virus-laden population accounted for 30% and 5% of exposed people, respectively, in the early and late stages of epidemic. The diagnosed probability of virus-laden population has increased from 7% to 40%, which shows that nucleic acid detection technology is becoming matured. Initial number of undiscovered people is about 70 000, and under effective national control, it is currently controlled at about 2 500. It is predicted that the “inflection point” will be mid-March, then residents can return to normal life at the end of April, and the cumulative diagnosis is about 100 000 finally.