Thread by @reichlab

Here is the weekly update from COVID Forecast Hub, where we are storing forecasts of #COVID19 deaths in the US from 20 research groups. Our national ensemble combines 7 models and predicts that we will see ~113K deaths by June 6 (80% PI: 104K-123K).
reichlab.io/covid19-foreca…

1/10

The ensemble model is now 85-90% certain that we will reach 100K #COVID19 reported deaths in the US by May 30.

This represents a slight increase and tightening in certainty compared with what the forecasts said last week.

https://twitter.com/reichlab/status/1257740028635222018

2/

We continue to see hints of increasing consensus among the models. Although the "best guess" predictions for total #COVID19 US deaths by early June range from ~103K to ~120K deaths. That difference represents a lot of people. More than sometimes die in a flu season.

3/

However at the state-level, particularly in states where it's not clear if the peak has arrived yet, the models still show quite large discrepancies. In PA, #COVID19 deaths by June 6 are forecasted by single models to be between ~4800 and ~9200.
4/

It's places like this where an ensemble model can help by providing a single "cone of uncertainty" about what will happen. In PA 3,779 deaths were observed as of Saturday. The ensemble predicts 6,839 deaths by June 6, with a 10% chance of having fewer than 6,168.

5/

At least 4 groups in the ensemble use information on current state-level social-distancing policies in their forecast model. Other models account for these policies implicitly, with assumptions about how the rate of new cases and deaths will change over coming weeks.

6/

If you are interested in digging into the details a bit further, all of the data, metadata, and code are public on GitHub:
github.com/reichlab/covid…

And our interactive, in-progress #dataviz is here:
reichlab.io/covid19-foreca…

7/

We continue to work closely with @CDCgov on this project, and versions of these data will appear on their #COVID19 forecasting site later this week.
cdc.gov/coronavirus/20…

8/

This has been a huge team effort to make this happen. Not just the teams for providing us their data, but also the COVID Forecast Hub team. Many of them not on twitter, but a few are including @johannesbracher and @jaradniemi

github.com/reichlab/covid…

9/

Many thanks to the 20 teams that have submitted their models and made their data public (partial list):
@meyerslab @covid_analytics @youyanggu @TAlexPerkins @alexvespi @sdelvall @sangeeta0312 @tomcm39 @IHME_UW @SenPei_CU @QuanquanGu @CovidActNow @sarahcobey @badityap

10/10