Epidemic Modelling with Parameters Updating based on Data-driven

When an unexpected epidemic outbreaks, a scientifically depicting of the evolution trajectory is a prerequisite for implementing various emergency response measures. Traditionally, parameters of the epidemic models, whether system dynamics models or time series regression models, are preset and fixed in advance. This poses the accuracy of forecasting declining with time. In this paper, we propose two modified epidemic models with parameters updating based on data-driven. The main contribution of this study is to continuously update the parameters with the actually collected epidemic data, in an effort to ensure that the forecasting result can be more close to the actual trajectory. Taking the Ebola outbreak in West Africa in 2014-2016 as our example. The test results demonstrate that the proposed two methods in this study are good at epidemic forecasting. Moreover, the modified system dynamics model is suitable for the medium and long-term forecasting while the modified ARIMA model is better at the short-term estimating.