Read on Twitter
Q: How hard is it to model covid cases in the future?
A: Very hard.
This CDC paper on modeling scenarios on vaccine efficacy/uptake and NPIs has gotten its share of Twitticule:
https://www.cdc.gov/mmwr/volumes/70/wr/mm7019e3.htm?s_cid=mm7019e3_w
It was well outside its 50% "prediction interval" before even being published!
A: Very hard.
This CDC paper on modeling scenarios on vaccine efficacy/uptake and NPIs has gotten its share of Twitticule:
https://www.cdc.gov/mmwr/volumes/70/wr/mm7019e3.htm?s_cid=mm7019e3_w
It was well outside its 50% "prediction interval" before even being published!
This is not inherently terrible.
Ppl who predicted the Chiefs would win the Super Bowl by >3 points were not fools;
they just underestimated how outmatched their OL would be.
But if I published a paper IN MARCH about how the Chiefs should've won the Super Bowl by more than 3...
Ppl who predicted the Chiefs would win the Super Bowl by >3 points were not fools;
they just underestimated how outmatched their OL would be.
But if I published a paper IN MARCH about how the Chiefs should've won the Super Bowl by more than 3...
I'd probably acknowledge the flawed parts of my analysis before advancing a model telling you to take the Chiefs at 6:1 to win the next Super Bowl.
Not much "acknowledging" happening with this paper.
Govts and institutions rely on models to make decisions for the future, now.
Not much "acknowledging" happening with this paper.
Govts and institutions rely on models to make decisions for the future, now.
So - I get asked a lot by patients, "What's going to happen with this pandemic?"
I have no idea, because I keep being surprised.
I did not see that little wave in late March'21 coming.
But if I had to guess...
I have no idea, because I keep being surprised.
I did not see that little wave in late March'21 coming.
But if I had to guess...
While it's clear that CDC ensemble model over-rated the case rise in March;
I fear they are far too optimistic for the rest.
And I cannot endorse that their projection interval gets smaller as the time gets further out!
I expect a push-pull between immunity, behavior&seasonality.
I fear they are far too optimistic for the rest.
And I cannot endorse that their projection interval gets smaller as the time gets further out!
I expect a push-pull between immunity, behavior&seasonality.
Summer brings on A/C gatherings.
Fall brings cooler, drier air and schools.
All the while, NPIs will be falling.
And maybe a variant like P.1.351 shakes up immunity and R0? Maybe vaccines peak w/in 6 months and natural immunity drops within a year, too?
So many variables
Fall brings cooler, drier air and schools.
All the while, NPIs will be falling.
And maybe a variant like P.1.351 shakes up immunity and R0? Maybe vaccines peak w/in 6 months and natural immunity drops within a year, too?
So many variables
I'd be very reluctant to predict very low levels through summer and fall.
I hope I'm wrong - but wouldn't be surprised to see cases >50K/d for long stretches between summer and next spring.
And... better to take the Steelers at 30:1.
Pls don't remind me of this thread next year.
I hope I'm wrong - but wouldn't be surprised to see cases >50K/d for long stretches between summer and next spring.
And... better to take the Steelers at 30:1.
Pls don't remind me of this thread next year.
