1/7 @brithume It makes no sense to credit distancing for the better numbers because the models needed to account for it! I can forgive if “models” from 4-6 weeks ago (when there was less US data) were way off and off regarding things 4-6 weeks away, but this isn’t the case. https://twitter.com/brithume/status/1247552520219656194

. @brithume The fact is, two weeks ago Monday, NY was expecting needing 30,000 ICU beds by this week. I showed at the time that it makes no sense. IHME’s models showed Thursday 3/26, that a week later NY will need 10,900 ICUs at a time when the total ICU count was a few thousand,

3/ and was going up by only 340 a day. IHME was essentially expecting the ICU count to pop one day to the next from an average of 340 into a daily average of 1,300. I tweeted at the time that it’s wrong. When 4/2 came around and NY was under 3,800, IHME simply pushed back

4/ the peak to this week Thursday and showed that there will be a need of 12,000 ICU beds despite having another week of ICUs growing by a fraction of what would be needed to actually pop from 3,700 to almost 12,000 in a week! This was the daily ICU rise in NY recently:

172 Fri 3/27/20

282 Sat

315 Sun

367 Mon

303

374 Wed 4/1/20

335 Thur

395 Fri

250 Sat

128 Sun 4/5/20

Despite this, IHME’s kept to its 11K - 12K projection, first for 4/2 and then for 4/9. No, not 5 weeks ago. Days away!

cc @brithume

5/7 As late as this week Sunday April 5th midday, IHME still projected NY will need almost 12,000 ICU beds despite the fact that that total ICU count in recent days was:

2,352 Sun 3/29/20

2,719 Mon

3,022 Tuesday

3,396 Wed 4/1/20

3,731 Thurs

4,126 Fri

4,376 Sat

6/7 Finally yesterday they cut it to NY needing 6,664 ICUs in a few days. Defenders of the models say “well, it takes time to update.” No! Not when the US is locked down due to you and not when NY, the state used by IHME to shape nationwide models, releases info daily!

7/7 If my calculator and Twitter account were able to show you weeks ago that you are way off, then you failed to update your own models. Models are not “scientific.” They are based on back-end data and assumption that humans enter or fail to enter. They failed! cc @brithume

My thread two weeks ago when NY expected needing 30,000 ICUs by this week: https://twitter.com/yossigestetner/status/1242938371141439488?s=21

Another thread from last week which addresses IHME later in the week

https://twitter.com/yossigestetner/status/1244413164873269248?s=21 https://twitter.com/yossigestetner/status/1242938371141439488