Games with Models, April 18 Data
I noticed something interesting today. Thus far, Governor Raimondo’s daily briefings on the coronavirus have produced a flurry of social media posts and articles emphasizing three metrics: cases, hospitalizations, and deaths. Over recent weeks, some journalists have said that hospitalizations is the key trend to watch, and when the governor gave a little bit of transparency into the data behind her decision-making, she released a chart projecting… yes… hospitalizations.
Well, while it may only be a temporary reprieve, the increase in hospitalizations has slowed way down in the past couple days, and suddenly that is no longer the subject of concern. Some local reporters have even switched to emphasizing only cases and deaths.
But that’s not how informing the public about trends is supposed to go. We shouldn’t follow a metric until it stops being interesting and then switch to something more topical.
So, to keep up the series of posts projecting hospitalizations in Rhode Island, here’s the chart based on today’s new data. Yesterday’s projection was for us to have 289 people in the hospital today, and the reported number was actually 255. It would be reasonable to suggest that the projection for tomorrow, of 293, is once again too high, but this model measures trends over the past week so as to be less volatile.
Based on the new numbers, the model suggests that tomorrow’s report will be:
- Cases: 4,755
- Hospitalizations: 293
- Deaths: 145
Unfortunately, although hospitalizations have been coming in below estimates, cases and deaths have been a little higher than projected.
Turning back to hospitalizations, the three models currently suggest they’ll peak as follows:
- RI Dept. of Health: From 255 to a peak of 2,250 (782% increase) in 15 days. (Even the governor is admitting this does not look likely, at this point.)
- IHME: From 255 to a peak of 610 (139% increase) in 5 days.
- Me: From 255 to a peak of 303 (19% increase) in 4 days.