Forecasting the Worldwide Spread of COVID-19 based on Logistic Model and SEIR Model

Xiang Zhou; Na Hong; Yingying Ma; Jie He; Huizhen Jiang; Chun Liu; Guangliang Shan; Longxiang Su; Weiguo Zhu; Yun Long

In Scenario 1, we assume that the epidemic trend obeys a logistic growth curve. We used a logistic model to predict the disease trends. The logistic model's essence is that curve fitting and its prediction results highly depend on historical data. It has often been used in the prediction of epidemic dynamics in previous studies 4, 10,11 . Mathematically, the logistic model describes the dynamic evolution of infected individuals being controlled by the growth rate and population capacity. According to the following ordinary differential equation (a), we will obtain the logistic function (b). The model describes the dynamic evolution of the reported number of confirmed cases P being controlled by the growth rate r, and the initial value of P 0 is the confirmed number of cases when T=0. The maximum case volume in the environment is K, which is the limit that can be reached by increasing to the final value of P ሺ t ሻ , and r is the growth rate. We used the least squares method to fit the logistic growth function and then to predict the number of future confirmed cases. Since the case numbers reported at very early stages are usually inaccurate or missing, the initial date of the model was set as the day since the 100th confirmed case was reached.