Modeling the COVID-19 pandemic - parameter identification and reliability of predictions

Hackl, Klaus

In this paper, we try to identify the parameters for two elementary epidemic models, the so-called SI- and SIS-models, via non-linear regression using data of the COVID-19 pandemic. This is done based on the data for the number of daily infections. Studying the history of predictions made, we attempt to estimate their reliability concerning the future course of the epidemic. We validate this procedure using data for the case numbers in China and South Korea. Then we apply it in order to find predictions for Germany, Italy and the United States. The results are encouraging, but no final judgment on the validity of the procedure can yet be made.