Alright, the problem I’m now having is how to:
A) Estimate the number of recovered cases, so that I can estimate the number of active cases. Data is way off state by state on recovered case reporting.
B) Factor in hospital discharges into their future capacity - in other words future recovered cases.
C) Fix my equations because it gets goofy with an R0 < 1. Like, Arizona has 0 to 11 days to hospital overrun. New York (R0 < 1) has 94 to 74 in my spreadsheet. Uhh, what? The best case scenario can’t be worse than the worst case! And this is also an issue with not accounting for recoveries. States that aren’t in a crunch and have an R0 < 1 will have increased available beds, not decreased.
I think the answer is to adjust the model to only take the last six weeks of data into account, and make it project remaining days using new cases - old cases, but I pulled an all nighter on poker and this, and can’t think well enough to make the changes right now.
That said, my R0 spreadsheet is now working, and I think my model should be reasonably accurate on the states that are close to tipover on their first major wave, since the recoveries and active case estimates don’t matter as much.
So here are those:
Arizona 0 to 11 days (tipover inevitability is reached when the back end < 12, but when I fix the recovery rate issue that could mean they haven’t quite collapsed the system yet, but Arizona is basically fucked as we know)
Florida 3 to 20 days (using the lady’s data who got fired, my model off the official data is 14-31)
South Carolina 9 to 22 days
North Carolina 15 to 38 days
Oregon 16 to 29 days
Vermont 16 to 26 days
Georgia 1 to 70 days (likely some issues with recovery messing this up by overstating active cases)
California 1 to 50 days (likely some issues with recovery messing this up by overstating active cases)
New Mexico -57 to 8 days (my guess is this is off due to many cases being concentrated on Native American reservations which likely have different hospital stats/access, and I can’t account for that in the model - also if my estimate is right anything on the high end that’s < 12 days is inevitably tipping over at this point)
It’s also important to note that, say, LA or Orange County could tip over while the state overall is still a long ways off from tipping over because my model is based on statewide hospital beds per capita.
The same applies to Dallas and Houston, which I’m estimating by hand off official hospital bed’s available and statewide R0 to be at 9 days and 8 days respectively, even though Texas overall is 46 to 74 days away. Both cities have a decent number of ICU’s and vents available, though, so they can probably scale their capacity as a result.
And here’s the R0 chart: