ELLIS against Covid-19

Inferring COVID-19 spreading rates and potential change points for case number forecasts

ellis 22 April 2020 - 22 April 2020
no image

22 April 2020 • 13:55 - 14:05

(Max Planck Institute for Dynamics and Self-Organization, Göttingen)

As COVID-19 is rapidly spreading across the globe, short-term modeling forecasts provide time-critical information for decisions on containment and mitigation strategies. A main challenge for short-term forecasts is the assessment of key epidemiological parameters and how they change as first governmental intervention measures are showing an effect. By combining an established epidemiological model with Bayesian inference, we analyze the time dependence of the effective growth rate of new infections. For the case of COVID-19 spreading in Germany, we detect change points in the effective growth rate that correlate well with the times of publicly announced interventions. Thereby, we can (a) quantify the effects of recent governmental measures to mitigating the disease spread, and (b) incorporate the corresponding change points to forecast future scenarios and case numbers. Our code is freely available and can be readily adapted to any country or region.

Video:

Question & Answers

Link to the recording of the live Questions & Discussion session for this talk. 

  • Q: The disease has a consistent incubation period (14 days 95-percentile). Why did you consider SIR and not SEIR?

    • A: We did also check other models, in particular SEIR-like implementations. The outcome is very similar with respect to the change-point detection. In terms of model comparison, the SEIR-like model did not describe the data better, however. You can check our arxiv paper for more details (Supplemental Material).

  • Q: The confirmed cases strongly depends on how much is tested... does your model include the capacity of the tests or something alike? Or is this implicitly assumed to be constant?

    • A: We do not include the test capacity in the presented model, but are currently working on implementations where this is included.

  • Q: I had a look at the preprint https://arxiv.org/pdf/2004.01105.pdf and I cannot see a plot (like the one you showed at the end of the presentation) where there is a weekly modulation of the number of infected cases, two different scenarios and the very large prediction intervals (from <1k to >8k cases) that you showed. Is that a new revision of the preprint? Can you explain why the PI are so large?

    • A: A similar plot to the one I showed at the end is Fig. S5.

  • Q: Why don't you put your code on Github in a public repository?

    • A: We did! The link was shown at the beginning and the end of the talk as a footnote. It is also Ref.31 in the arxiv preprint. Here it is again: https://github.com/Priesemann-Group/covid19_inference_forecast

Speaker(s):

Thumb ticker johannes zierenberg
(Max Planck Institute for Dynamics and Self-Organization, Göttingen)