An empirical forecasting method for epidemic outbreaks with application to Covid-19

Abstract

In this paper we describe an empirical forecasting method for epidemic outbreaks. It is an iterative process to find possible parameter values for epidemic models to best fit real data. As a demonstration of principle, we used the logistic model, the simplest model in epidemiology, for an experiment of live forecasting. Short-term forecasts can last to 5 or more days with relative errors consistently kept blow 5%. The method should improve with more realistic models.