Abstract:
The article presents model based forecasts of daily confirmed new cases with Covid-19 in Norway and for the number of hospital beds. The model, CovidMod, is complementary to the mathematical epidemiological models used by NPIH to forecast during the pandemic. CovidMod is an empirical mathematical model. It is the product of the statistical modelling of historical realizations of the time series that the goal has been to forecast. A comparison with the 21 day ahead forecasts from the NIPH’s Regional National Model shows that CovidMod forecasts had lower RMSFE (Root Mean Squared Forecast Error), both for new cases and for hospital beds. The large wave in new infections in November and December 2021 is in itself an interesting case. In the model, this wave is explained by the coming of the Delta version of the virus and by the partial lifting of the non-pharmaceutical measures earlier in the autumn. The joint effect of these factors was to lift the model's critical parameter from the stable to the explosive region. In the model this happened despite the increase in the vaccination rate, which has an estimated negative effect of hospitalization in particular.