Games with Models, 5/26/20


After a long, Memorial Day weekend, we return to good new on Rhode Island’s COVID-19 front.  Despite some loosening of the lock-down and people’s patience with the rules wearing thin, for two days, new positive cases have been under 80.  Granted many fewer people were tested, but that could be because fewer people felt the need to be tested.

As the chart below shows, hospitalizations are down, too, more or less in line with projections.  It’s worth noting, again, that the hospitalization numbers are two days behind and that the drop in hospitalizations would be lower but for a reduced number of discharges.

The number of daily deaths appears to provide some room for hope, too.  Although they’ve been subject to periodic revision, it appears at the moment that we’ve shifted to a lower gear, with deaths consistently in the single digits each day.



(See here for my original methodology and here for a subsequent modification I made.)

Projections versus actuals (date of report).

  • Cases:
    • Projection for 5/26: none made
    • Actual for 5/26: 14,210
    • Projection for 5/27: 14,314
  • Hospitalizations
    • Projection for 5/26: none made
    • Actual for 5/26: 226
    • Projection for 5/27: 219
  • Deaths
    • Projection for 5/26: none made
    • Actual for 5/26: 634
    • Projection for 5/27: 639



  • Mario

    I’m not sure I’m as optimistic as I was. I still showing it ending roughly the same time, but now it’s early July with over 800 deaths instead of late June at 700. The fact that the date and severity keeps getting pushed back is a good indication that I was wrong about the timeframe entirely. I have said it a lot when it comes to the national numbers, but this isn’t really a single infection, it’s a group of clusters. Dividing all of the numbers over the total population makes everything look less severe than it is. I hadn’t even considered what that meant locally, but during the press conference today they said something like 20 or 25% of the tests in the urban areas are coming back positive. I should have considered that. 5000 infected people spread across the state is something that can be eliminated the way I expected, but if the infections are a lot clumpier, we are dealing with a different problem. It’s like radiation, it gets worse as it’s concentrated.

    I’m no longer convinced that the infected population is slowly shrinking, I think it might even be growing again, possibly in places that aren’t being well-covered by testing. I think we’ll see eight deaths, 184 cases, 218 hospitalizations (+19, -23, -3).