This study demonstrates that non-identifiability in model calibrations using confirmed-case data is the main reason for such wide variations in existing model predictions of the COVID-19 epidemic in China.

It is crucial for modelers to estimate the severity of the COVID-19 epidemic in terms of the total number of infected, total number of confirmed cases, total deaths, and the basic reproduction number, and to predict the time course of the epidemic, the arrival of its peak time, and total duration. Such information is needed to help the public health agencies make informed decisions.

But since the COVID-19 outbreak in Wuhan City in December of 2019, numerous model predictions on the COVID-19 epidemics in Wuhan and other parts of China have been reported, with wide variation.

The authors of this study present both SEIR and SIR model predictions for the COVID-19 epidemic in Wuhan after the lockdown and quarantine of the city on January 23, 2020. Using the Akaike Information Criterion (AIC) for model selection, they show that an SIR model performs much better than an SEIR model in representing the information contained in the confirmed-case data. Their findings indicate that predictions using more complex models may not be more reliable compared to using a simpler model.