COVID-19 Event Risk Assessment
Based on the model of the Georgia Institute of Technology.
Base number: Positive tests in the past X days per 100k population. CDC recommends to use past 10 days.
As there is no recent serology data about Switzerland, it's hard to estimate how many people are infected, but untested.
One large seroprevalence study was performed in Geneva in April. The results showed that for one case tested, there were 11.8 untestet.
This would mean a bias of 11.8 - but the test regime was different during these times and it's difficult to set the correct value.
Also a study in Ticino performed in May resulted roughly in a 1:10 ratio.
As default, the value of three is set, which is on the lower end I assume - but I am just a programmer.
For certain categories of events (schools, nightclubs), it would be very useful to filter by age group,
as the distribution of the virus varies greatly between the groups. This is not implemented yet.
Formula used: 100 - (((1-(1/(100000/(cases*bias)))) ** groupsize) * 100)
Attention: be aware, don't missuse it! Don't use it for your personal risk assessment!
"Oh, the chance that someone is infected at this party is only 9,3%? Let's go and hug everybody!"
Probabilities are tricky to understand - this tool is maybe more useful for decision makers/event organizers (keyword "Superspreader in closed space").
Disclaimer: I am not an epidemologist. I'm just a random dude from the internet.
Credit for the concept goes to Gatech, I just adapted their concept for Switzerland.
So: keep your distance, wash your hands, wear a mask where it's sensible, enjoy life but don't be stupid.
Cases per 100k population in the past...
Ascertainment bias: incidence is multiplied by this number.