Games with Models, 5/4/20 Data


Rhode Island’s daily reports on COVID-19 are settling into a pattern over the past few days.  The number of total cases is growing more slowly each day, and hospitalizations are generally down, while deaths continue to increase at a more or less steady pace.  The latter two measures, meanwhile, are regularly revised going back weeks, making it difficult to know what the trends actually are.

The best news of today’s report is that new cases saw its smallest one-day increase since April 7, with only 175 positive cases despite an increase in testing compared with the day before.  (We should note, by the way, that a reduction in the number of tests isn’t necessarily evidence of failure, but could indicate that fewer people need to be tested.)

On the downside, the number of people in the hospital increased, but as previously discussed, we can no longer attribute those hospitalizations to the disease itself.  That is, we are no longer being told who is in the hospital because of COVID-19, but rather, the number represents the number of people in the hospital while having COVID-19, even if they are completely asymptomatic.  (Of course, compared with the state’s original projections that we’d have thousands of people in the hospital because of COVID-19 right now, the gaps between the different was of counting the actual data seem like quibbling.)

My updated projection graph follows:



Just for fun, I’ve added the line from the above chart to the state’s original projection, which led the governor to take such extreme measures, and the rest of us to accept them.  My line is the red one way down low.


And here’s how my projections compared with reality today:

  • Cases:
    • Projection for 5/4: 9,633
    • Actual for 5/4: 9,652
    • Projection for 5/5: 9,805
  • Hospitalizations:
    • Projection for 5/4: 315
    • Actual for 5/4: 339
    • Projection for 5/5: 323
  • Deaths:
    • Projection for 5/4: 325
    • Actual for 5/4: 341
    • Projection for 5/5: 346

  • Mario

    I was wrong about how low the new case number could go, at least as much due to the appallingly low number of tests as the good news of the sub-8% positive rate. I was critical of what your model said for yesterday, but it is very plausible for tomorrow, largely because last Tuesday’s testing was so bad, but this is an area where I long for some “bad” news. A lot of people have the virus anyway, a low number of new cases just means that some of those people aren’t being found. Now that’s we’re at least temporarily on the down slope, I want that number to be as high as possible. The higher the number of new cases, the faster the recovery. They announced today that they’ll be testing more broadly, and while that’s nice, I thought this is what was happening two Saturdays ago. It’s so late.

    Hospitalizations: 341 (+23, -15, -6), a small increase, but I expected the same for today and I was right about that. It just seems to be a temporary dip in the number of discharges. Deaths I have at 8; I keep coming up short, but I do think a slight correction is due. We’re definitely not at the 5-6 deaths per day I was hoping for a week ago. I can blame that somewhat on all of the revisions, but this is all so much uglier than it looked like it would be. I have 30 deaths coming over the next 3 days, but it could easily be over 50 (and that’s not counting more revisions). After that it should start dropping a bit more.