Big Finding By Northeastern University

[ProfessorGAC]

It's possible, per this study, that 28,000 people were infected when the official count was 23.
That's 23, with no zeroes.
This is also being discussed on CNN.
I'm dubious of CV models right now, but if we're going to keep examining the IMHE model, probably should at least know about this one.

https://thehill.com/policy/healthcare/494258-coronavirus-spread-undetected-before-testing-showed-problems-researchers

[pwb]

1. I cant put my head around Atlanta being a transportation hub

and having a low number of cases? All the other hubs are very high.

[ProfessorGAC]

2. Seems This Is Cementing The Notion...

...that we aren't close to enough testing!

[Igel]

7. Different kind of hub.

Transmission means a lot of people pass through and stay. Look at JFK airport. People land and disembark for NYC, their destination; some transfer to Newark, and mix a different way in transit. Even a lot of people heading from Paris to Los Angeles with a layover in NYC opt to make that a 48-hour layover instead of 2 hours, so it's an interim destination not just a place to be for an hour or two in the terminal.

But if you are heading from Paris to Los Angeles you rub elbows with a lot of people who rub elbows with locals.

Now, Atlanta. It's a hub--like Houston. Lots of people go through, but you know, if I have to transfer at Atlanta I'm not going to be tempted to take an extra couple of days. And not that many people exit to Atlanta.

When you look at the labels, think of what's behind the labels. If the virus requires human-to-human transmission and usually air-borne particles, look for human-to-human contacts in close quarters and where you're coming into contact with a lot of different humans.

Some airports you go through and there's a lot of room. Planes are mostly on time. Others, they're cramped and stuffy. I haven't been through JFK, I don't think, and can't say I remember Newark's.

[Amishman]

3. The models are sketchy, and the data they are being fed doubly so

Lets says something like this. Antibody test X has a 10% error rate (which is very possible, some are thought to be worse).

Test 200 at completely random, lets say 5 actually have it, 195 don't.

apply 10% error rate and you get results saying 24 have it, or 12%. That 10% error rate applied to a low infection population results in overreporting of infections by more than double.

Put this into a flawed model, and the output is garbage.

We don't know anything with any real certainty.

[ProfessorGAC]

4. I Think I Said That

You seem to be arguing while agreeing.

[central scrutinizer]

5. Silver lining

If the higher number is accurate then that dramatically changes the denominator in the fatality rate calculation. That’s over three orders of magnitude! Of course, the numerator is probably not correct either if some fatalities were never tested

[ProfessorGAC]

6. Good Thought

It's hard to believe, given the paucity of testing that the #infected is not much higher than reported.

[JCMach1]

8. You can't model anything without proper R0

We are operating blind because of lack of testing

[ProfessorGAC]

9. I'm Debating Nothing JC

There is now a new model based upon sound computational science.
I never said we should blindly trust the results. It's just new information.

[JCMach1]

10. I know... The info in the OP is valuable because

It's just becoming clear how much we are flying blind

IMHE and other models must have an accurate R0, or they are pretty useless for making policy

[ProfessorGAC]

11. Got Ya!!

I saw the guy from U of I today on Pritzker's briefing. They've assembled a group of epidemiologists & computational experts to actually model the models. Three big universities are involved.
They said they were trying to interactively coalesce 3 different models to "come up with predictions on predictions".
Good luck with that!
I have to say, he was impressive as hell!