Bad News Thread

N00B said:
What variable are you currently using for asymptomatic or non-test confirmed COVID cases? At one point assumptions ranged up to 10x, but I’m not sure what the current assumption should be.

In September models were using 9x, but I wonder where it is at now?

www.nature.com

Substantial underestimation of SARS-CoV-2 infection in the United States - Nature Communications

Estimating the extent of SARS-CoV-2 infection in a population is challenging due to the limitations of testing. Here, the authors estimate that the true number of infections in the United States in mid-April was up to 20 times higher than the number of confirmed cases.
www.nature.com www.nature.com
Click to expand...
Can't be 9x. Some places are above 12%. That gives you a 108% infection rate.

An OC study in the fall found 6x, for them..

Even 6x gives you above 100% infection rate for certain counties.

Going with 4 or 5x, depending on positivity rate. High positivity implies higher multiplier.
 
dad4 said:
Can't be 9x. Some places are above 12%. That gives you a 108% infection rate.

An OC study in the fall found 6x, for them..

Even 6x gives you above 100% infection rate for certain counties.

Going with 4 or 5x, depending on positivity rate. High positivity implies higher multiplier.
Click to expand...

Thanks.
 
Saw a report today that Kaiser has said if vaccine production/distribution doesn't vastly speed up, it take them four years to vaccinate all their members.

I'm not one of the smart ones working on this stuff but it seems to me that it would cost the US less to throw ALL of it's resources to fixing that than it's going to cost them indirectly from all the jobs being lost.
 
Glitterhater said:
Saw a report today that Kaiser has said if vaccine production/distribution doesn't vastly speed up, it take them four years to vaccinate all their members.

I'm not one of the smart ones working on this stuff but it seems to me that it would cost the US less to throw ALL of it's resources to fixing that than it's going to cost them indirectly from all the jobs being lost.
Click to expand...

If the FDA were to be more efficient and approve the J&J and AZ vaccines (the AZ has already been approved in the UK and IIUC Europe), it would quickly double our available supply particularly because of the need for only 1 dose in the J&J vaccine. The problem is a lower overall efficiency, which isn't good enough for some of the most vulnerable, but should be fine for all adults less than elderly or severely disabled.
 
Glitterhater said:
Saw a report today that Kaiser has said if vaccine production/distribution doesn't vastly speed up, it take them four years to vaccinate all their members.

I'm not one of the smart ones working on this stuff but it seems to me that it would cost the US less to throw ALL of it's resources to fixing that than it's going to cost them indirectly from all the jobs being lost.
Click to expand...
Nationally, we are at about 5%. 8% first dose, 2% second.

Annoyingly slow, but I don't see 4 years. NYT puts it at December, based on current rate. Should speed up if/when they approve J&J or AZ.
 
dad4 said:
Can't be 9x. Some places are above 12%. That gives you a 108% infection rate.

An OC study in the fall found 6x, for them..

Even 6x gives you above 100% infection rate for certain counties.

Going with 4 or 5x, depending on positivity rate. High positivity implies higher multiplier.
Click to expand...

So as the variable for ‘undocumented’ (test positive confirmed cases) shifts based on increased testing, positivity rate, etc., is it appropriate to run a regression based on a curve over time as the variable changes in order to create a predictive model (polynominal)? Most models or assumptions regarding heard immunity I’ve seen seem to be linear.
 
N00B said:
So as the variable for ‘undocumented’ (test positive confirmed cases) shifts based on increased testing, positivity rate, etc., is it appropriate to run a regression based on a curve over time as the variable changes in order to create a predictive model (polynominal)? Most models or assumptions regarding heard immunity I’ve seen seem to be linear.
Click to expand...
you could do normal Runge-Kutta, but it gets wonky as you go further out in time and the projection gets further from the data. You’d also have to use smoothed data, since you don’t want a single weird day (Christmas?) throwing a monkey wrench into your highest order term.

I’ve been using a declining exponential model for any region that is post peak. Seems good so far. I tweak it a bit since increasing vaccinations increase the rate of decline.

I think it will start to be overly optimistic as numbers fall. People will start relaxing behavior, and the decline will be slower than I predict. No idea how to model that.

At the moment, still thinking late March for orange. Worse to the extent that we all hit the casinos and host dinner parties.
 
dad4 said:
you could do normal Runge-Kutta, but it gets wonky as you go further out in time and the projection gets further from the data. You’d also have to use smoothed data, since you don’t want a single weird day (Christmas?) throwing a monkey wrench into your highest order term.

I’ve been using a declining exponential model for any region that is post peak. Seems good so far. I tweak it a bit since increasing vaccinations increase the rate of decline.

I think it will start to be overly optimistic as numbers fall. People will start relaxing behavior, and the decline will be slower than I predict. No idea how to model that.

At the moment, still thinking late March for orange. Worse to the extent that we all hit the casinos and host dinner parties.
Click to expand...

Exhibit A as to why @Grace T. *secret handshake* respects @dad4 math skills.
 
dad4 said:
you could do normal Runge-Kutta, but it gets wonky as you go further out in time and the projection gets further from the data. You’d also have to use smoothed data, since you don’t want a single weird day (Christmas?) throwing a monkey wrench into your highest order term.

I’ve been using a declining exponential model for any region that is post peak. Seems good so far. I tweak it a bit since increasing vaccinations increase the rate of decline.

I think it will start to be overly optimistic as numbers fall. People will start relaxing behavior, and the decline will be slower than I predict. No idea how to model that.

At the moment, still thinking late March for orange. Worse to the extent that we all hit the casinos and host dinner parties.
Click to expand...
One concern I have is that those in geographic areas with a lower "R" are getting the vaccine at a higher rate than those in areas with a higher "R". Models that assume vaccines are going into the "average" risk for a given group may end up being overly optimistic.
 
kickingandscreaming said:
One concern I have is that those in geographic areas with a lower "R" are getting the vaccine at a higher rate than those in areas with a higher "R". Models that assume vaccines are going into the "average" risk for a given group may end up being overly optimistic.
Click to expand...
SCC started a walk in vaccine clinic in the middle of the bad zip codes. And some places are doing giant drive through clinics. So some hope for covering the hard hit areas.

Bad vaccine targeting does not hit the model too hard. The vaccine term is still a very small part of overall immunity, and will be the minor player well into summer. Also, hard hit areas are getting a smaller slice of the vaccine immunity, but they already have a larger slice of traditional immunity. If we tie our hands as we vaccinate, it shows up as a lower vax rate and higher cases. So the model is ok.

The people? Maybe not. All the special categories and phone trees just make it harder for people in high risk areas to get vaccinated. A simple system by age would be both fairer and more efficient.
 
I was talking to a physician friend last night, he's actually in New York City, So he has seen a lot!

Anyway, he somewhat echoed the statement that this thing is never going away. That the only way to get it to where we can function on a regular level is with the vaccine- Then it's going to be looked at as the flu is looked at. He said in his opinion, we are always going to be looking over our shoulder with all these variants until the vaccine is widespread.

If I remember correctly, I think Grace was one of the ones saying this a few months ago? Pardon me if my memory is fuzzy.

Something I forgot to add: he said he's no more concerned about the UK variant than he is any other variant- meaning, they are all trouble for us.
 
Grace T. said:
Yeah. particularly since the data seems to show we are in a race against a 3rd wave.
Click to expand...

On a linear scale, it looks like we have already had 5 waves, but the early ones get swamped out by the size of the later ones on a logarithmic scale.

aatishb.com

Covid Trends

Visualizing the exponential growth of COVID-19 across the world.
aatishb.com aatishb.com
 
As much of a fringe group as the COVID deniers, it's interesting that some people still believe we can get to zero COVID. What's more troubling is that there may be a coming clash of certain governments (the PRC, Singapore, Taiwan, South Korea) that want to eliminate COVID v. other governments that just accept we can't.

unherd.com

Inside the Zero Covid campaign

The urge to eliminate the virus is understandable — but at what cost?
unherd.com unherd.com
 
You must log in or register to reply here.
Back
Top