The forecast of 58,000-305,000 more deaths from COVID-19 in the U.S. by mid-March, circulated by an AP article, is not based upon one computer model, but actually the average of various, different computer models. Katriona Shea at Pennsylvania State University is the co-lead of a team that reviews various models and reports the forecasts to the White House. Their average forecast puts the number of deaths at 191,000 for the mid-December to mid-March period, with low and high estimates at 58,000 and 305,000. Given that there’s already been over 50,000 such deaths since mid-December, it is safe to say that the more conservative models have already been shown to be off target.
Further, given the current rate of around 1,900 deaths/day, another 114,000 deaths would be added by mid-March, for a total of over 164,000—and the 1,900 deaths/day rate is still increasing. (The 191,000 figure provided translates to a bit over 2,100/day for the whole 90 days.) Further, the estimations of the various models all seem to assume: 1) that the number of new cases will peak in early February and then recede, leading to a drop in the deaths; and also 2) that new variants will not be generated out of the present surge.
