=> numbers last updated: Thu, 04 Feb 2021 02:54:17 CET <=

During the last 6 weeks, we provided COVID-19 case number forecasts on this page, fitting a containment model to the total number of confirmed cases for many countries. From now on, we will not provide any further updates.

The reasoning behind this: Many countries implemented strategies to contain the further spread of the disease, which we argue is reflected in a slowed growth of cases over time. In our model, this effect is explained by assuming a constant containment rate, with which susceptibles are successively removed from the transmission process. The model assumes that asymptotically all susceptibles will be removed from the transmission process by either infection or containment, which results in a temporal reproduction number that approaches zero asymptotically. Naturally, this assumption is violated for longer times in reality: Not everybody can and will remain completely isolated for an extended period of time. Rather, one may assume that after reaching a peak in new infections per day, a saturation behavior sets in where the temporal reproduction number reaches a non-zero value, or even increases again after containment measures are lifted. Therefore, our model is prone to systematically underestimate the total number of confirmed cases after a peak was reached. Further providing updates with our model might lead to a false sense of security to which we do not want to contribute to. Hence, we decided to refrain from updating the case number fits from now on. We are currently working on model variations that assume that sanctions are slowly lifted and/or that containment efforts yield a saturated, non-zero level of the temporal reproduction number. This, however, is new research, and doing it properly needs time. We apologize for any inconvenience this might cause.

Forecasts by Country

On this page, we present the 6-day forecasts of COVID-19 case counts by country based on a novel epidemiological model that integrates the effect of population behavior changes due to government measures and social distancing.

The SIR-X model is described in detail here: Effective containment explains sub-exponential growth in confirmed cases of recent COVID-19 outbreak in China, B. F. Maier & D. Brockmann, Science, eabb4557, DOI: 10.1126/science.abb4557, (2020)

The containment measures implemented in response to the growing pandemic vary drastically by country. Classical epidemiological models fail to capture the impact of such efforts on the spread of the outbreak. Under unconstrained conditions, we would see exponential growth in the number of confirmed cases. However, several graphs below indicate that this is not the case. These insights can be used to evaluate the effectiveness of containment strategies in order to inform further courses of action and future policies.

Click a country below to view the forecasts for that country. Move the pointer to display the number of confirmed cases by date.

The open dots indicate the total number of confirmed cases over time. The blue bars represent the new confirmed cases per day. The solid line depict the model’s fit and subsequent predictions of case count numbers for the next 6 days as well as the expected new cases per day. The grey and red shaded regions represent the 98% and 68% confidence intervals, respectively.

Please note that the model assumes that the number of contacts between individuals can change over time continuously by susceptibles going into isolation or otherwise being shielded from the transmission process. Discontinuous, disruptive changes are not captured by the model. For countries where abrupt changes can be seen in the data, it is likely that infrastructural changes play a role in how cases are counted. The model also puts higher weights on most recent data points to capture the latest development accurately.

You can read more about the model, its assumptions and the accuracy of its predictions on another page.

All case data is provided by Johns Hopkins University’s Center for Systems Science and Engineering (CSSE) via their online GIS dashboard.

Reference