gabgoh / epcalc Goto Github PK

Nice calculator! It seems the label "case fatality rate" slider should be changed to be "infection fatality rate"? If you're going to use CFR, you would need an extra variable or two to specify what fraction of infections are diagnosed, and/or model the testing process itself.

Hi,

Your calculation for infections:
dI = a*E - gamma*I # where gamma = 1 / D_infectious
However, D_infectious is defined as the days someone is contagious. This seems unrelated to the number of people that recover from infection.

According to the wiki page for SEIR, gamma should represent the recovery rate:
gamma = 1 / days_to_recover

Am I understanding the code/wiki wrong?

Erwin

Hey,
could you add a license? I'd love to embed this into some internal views etc

Hi @gabgoh,

Congrats of this app! A lot o people think it is fantastic.

Would there be possible to export the list of input parameters and the resulted projections in a .csv file in order to be included in reports?

Thanks a lot!
Marius

Would it be possible to add an option to type the numbers in the calculator in addition to the sliders?

Hi there Gabgoh,

I haven't used JavaScript in 15 years or so, but I have now created a fork of your repository trying to include the country specific real world data from John Hopkins University in the visualisation.

So far I am able to parse the github data for a specific country / region (the parsing code is in my fork of Chart.svelte). If you also think that this might be an idea worth exploring I would be more than happy to join forces. It may also be a good idea to develop this as a separate project. However, on the chance of being useful I wanted to share the (ongoing) work early.

Most public health agencies are publishing "total hospitalized till date" data as opposed to currently hospitalized. To tune the model parameters it would be very helpful for the calculator to publish "total hospitalized" in addition to currently hospitalized.

Hey, I really love this calculator and I think this is an amazing piece of work! I noticed this minor issue and I would be happy to provide a PR for it if you'd be interested?

I noticed that the generation of values is tied to the presentation of values (to be specific, it's tied to the timestep/zoom of the graph). For example, using the default parameters and default zoom, day 100 infected count is 66,006. But if you zoom in on the graph so that day 100 is the last day shown, the infected count changes and is now 66,977. This appears to happen because, although the label says "day 100", it might actually be "day 100.38" (or whatever, depending on the timestep).

I think generation of values should be entirely separated from the presentation of values: generate values for all the days and then show values for the desired days. With this change the predictions don't change depending on the zoom. This change would also require tweaking the way zooming works on the graph.