SPI-M-O: COVID-19 short-term forecasts, 17 May 2020
Short-term forecasts prepared by the Scientific Pandemic Influenza Group on Modelling, Operational sub-group (SPI-M-O).
Documents
Details
Short-term forecasts from SPI-M-O. This was considered at SAGE 37 on 19 May 2020.
It should be viewed in context: these short-term forecasts represent SPI-M-O’s best predictions based on the data and evidence available at the time of writing. Therefore, some of the information in this paper may have been superseded and the author’s opinion or conclusion may since have developed.
Separate forecasts are produced using different models and approaches by the modelling groups represented at SPI-M-O. These individual forecasts are then combined to form a consensus forecast which is subsequently shared with the Scientific Advisory Group for Emergencies (SAGE).
The short-term forecasts are produced by transmission models of the epidemic process. These models are fit to trends in the past data, and therefore do not predict the impact changes to social distancing measures will have on the number of COVID-19 deaths or hospitalisations - unless the impact of any changes have already started to be reflected in the data by the time the forecasts were produced.
It isn’t possible to produce accurate forecasts when case numbers fall to very low levels. Furthermore, where data series are inconsistent, for example if ICU bed occupancy is decreasing much faster than general bed occupancy, the models may not always fit well to the data. If case numbers drop to very low levels in a region or the forecasts do not fit well to the data, then these forecasts are excluded from the slides circulated to SAGE.
Data limitations mean some of the forecasts aren’t informed by past data. Where this is the case these forecasts need to be treated with caution and represent SPI-M-O’s best assessment using the available data from other parts of the UK.
These documents are released as pre-print publications that have provided the government with rapid evidence during an emergency. These documents have not been peer-reviewed and there is no restriction on authors submitting and publishing this evidence in peer-reviewed journals.