An influential coronavirus model nearly doubled its US death toll prediction overnight

An influential coronavirus model — one often cited by the White House coronavirus task force — nearly doubled its United States death toll estimate in a revision to its projections Monday.

The Institute for Health Metrics and Evaluation at the University of Washington now predicts that 134,475 Americans will die as a result of the virus by Aug. 4. Its previous April 29 release, which ran unchanged until Monday, predicted that 72,433 Americans would die during the same time period — that's an increase of 62,042 deaths.

What changed?

According to the update notes from the new release, the higher projections take into account new data on the increased mobility of Americans and the loosening of social distancing guidelines.

"These projections are considerably higher than previous estimates, representing the combined effects of death model updates and formally incorporating the effect of changes in mobility and social distancing policies into transmission dynamics," the notes say.

It's true that data has emerged in recent days showing that Americans are no longer sheltering in place to the extent they had been in early April, despite continued social distancing guidelines in many states.

The update notes also highlighted specific states such as New York, New Jersey, and Pennsylvania, where the modeling significantly increased its cumulative death totals.

The higher predictions, researchers wrote, is based on "longer epidemic peaks and slower downward trajectories" in those areas.

Some disclaimers

TheBlaze has been reporting on coronavirus models for some time, giving specific attention to the IHME model, which has emerged as one of the key models used by the government for predicting U.S. deaths and hospitalizations.

While several early models forecasted grim death toll predictions as high as 2.2 million, the IHME model, has been comparatively modest in regard to death toll projections. In its first release on March 26, researchers predicted that 81,000 Americans would die from the virus. In subsequent revisions, the number had always landed under the 100,000 number, until Monday.

However, though modest, the model has certainly been erratic, constantly altering its projections by thousands at a time.

Moreover, besides death toll projections, the model also predicts state-by-state hospitalization numbers which state and local officials use to prepare for the outbreak in their jurisdictions. These estimates have often been wildly inaccurate, failing to accurately predict hospitalizations for the very next day, never mind months down the road.

In the new May 4 release, IHME notes that it uses The COVID Tracking Project to gather state-by-state data. That's the data tracker that TheBlaze used to assess IHME's hospitalization predictions. It should be noted that the project's data is only as reliable as the data given to it by individual states.