Discussion
This study compared the predicted number of deaths in each US state as of May 5, 2020 to the number predicted on March 30, 2020 for four months later (July 30, 2020). The findings show that, as expected, in most states the actual number of deaths on May 5 was lower than the predicted number for July 30 (almost three months later). Nationally, there were about 77% as many deaths in the United States on May 5 as the IHME predicted there would be on July 30.
However, the accuracy of the predictions was highly variable across states, and of the 51 regions (50 states and the District of Columbia) included here, 8 of them had already surpassed the predicted number of deaths nearly three months in advance. Furthermore, extrapolating the number of new deaths per day in the 36 days between the IHME's publication of their article to the writing of the present paper until July 30 leads to the prediction of 213,689 deaths in the United States by that date, more than double the IHME's prediction of 81,111 deaths. Hence, these results corroborate other arguments that the IHME's predictions for the United States are overly optimistic.\cite{kelsey2020}\cite{adam2020}