Games with Models, April 17 Data

COVID19-hospitalizationsandprojections-041720-featured

It occurs to me that daily comparisons of COVID-19 predictions versus reported numbers would be more useful if the model is adjusted soon after the daily data dump, so I’ll put up two of these posts today.  This one adjusts my model for yesterday’s numbers, and this afternoon, I’ll put up another based on whatever news the governor reports.

For April 17, I had predicted 282 total hospitalizations (as in, people currently in the hospital).  The reported number was 252.  Adjusting for the reported numbers of cases and hospitalizations produces the following:

COVID19-hospitalizationsandprojections-041720

 

For the record, the underlying predictions that produce this line suggest that today the governor will announce:

  • 4,475 cases
  • 289 hospitalizations
  • 126 deaths

As of this writing, the three models I’m tracking suggest the following change in hospitalizations:

  • RI Dept. of Health: From 252 to a peak of 2,250 (818% increase) in 16 days.
  • IHME: From 252 to a peak of 610 (142% increase) in 5 days. (Note that this has changed dramatically since yesterday, when the IHME was predicting a peak closer to the governor’s date.)
  • Me: From 252 to a peak of 313 (24% increase) in 4 days.