Model of COVID-19 spread for Germany using data form Jan-22 to March 30, 2020.
Methods[edit | edit source]
I have produced a projection on COVID-19 spread for Germany based on
(1) a logistic growth model
(2) fitted to the Johns Hopkins University data on known/reported infection cases.
The calculation was done wth SPSS' non-linear regression. Testing the method on the data from Hubei Province/Wuhan gave a model fit (R2) of 99,5%. Fitting the German data from Jan-22 to March-30 gave a model fit of 99,9%.
Results[edit | edit source]
Main result: By end of the end of April 2020, the maximum of the current (March 30, 2020) epidemic wave may effectally be reached with an estimated 102,000 cases of reported infections. The 95% confidence interval rages from ~95,000 to 109,000 cases.
Caveats[edit | edit source]
- validity of results depends on long list of conditions among which are
- logistic growth model works sufficiently well to describe COVID-19 epidemic in Germany
- Wuhan/Hubei data fit was 99,5 %
- validity of results depends on validity of the data used
- projected maximum is maximum of known cases; actual number of infections is much higher
- a new infection mechanism may show up that is not reflected in the data up to March 30, 2020
- Here, we model only the first wave of infections. Later-on, 60-80 % of the entire population may still get infected.
Attention![edit | edit source]
This calculation may be completely wrong in the sense that the next few weeks play out completely differently!
The results implicitly assume that people do NOT change their protective behavior based on these (or other) projections that give some hope.
If people would take it more easily, the projection is guaranteed to be too optimistic.
Best read as a best case scenario, in which no new sources of infections show up and people continue to self-isolate, wash their hands, … , and strictly follow public health rules.
Files[edit | edit source]
File:COVID19-31-03-2020srt.pdf - More detailed explanation of methods and results