A data-driven model for predicting the course of COVID-19 epidemic with applications for China, Korea, Italy, Germany, Spain, UK and USA

Huang, Norden E; Qiao, Fangli; Tung, Ka-Kit

For an emergent disease, such as Covid-19, with no past epidemiological data to guide models, modelers struggle to make predictions of the course of the epidemic (Cyranoski, Nature News 18 February 2020). The wildly varying predictions make it difficult to base policy decisions on. On the other hand much empirical information is already contained in data of evolving epidemiological profiles. We offer an additional tool, based on general theoretical principles and validated with data, for tracking the turning points, peak and accumulated case numbers of infected and recovered for an epidemic, and to predict its course. Ability to predict the turning points and the end of epidemic is of crucial importance for fighting the epidemic and planning for a return to normalcy. The accuracy of the prediction of the peaks of the epidemic is validated using data in different regions in China showing the effects of different levels of quarantine. The validated tool can be applied to other countries where Covid-19 has spread, and generally to future epidemics. US is found to have the largest net infection rate, and is predicted to have the largest total infected cases (708K) and will take two weeks longer than Wuhan to reach its turning point, and one week longer than Italy and Germany.