Why Do We Have Data Problem?

4,027 Views | 114 Replies | Last: 2 mo ago by Wife of Chas Satterfield
Agnzona
How long do you want to ignore this user?
Why don't we have good data? Why can't we run real solid models?

By now we should have a very solid death rate and should be able to extrapolate, deaths based on our actions. i.e. shelter in place, work with social distancing, complete return to normal activity???
Rendered Fat
How long do you want to ignore this user?
It wouldn't matter anyway. Bo Darville, InfectionAg and P.C. Principal are the only ones who understand basic math.
GMaster0
How long do you want to ignore this user?
Read the NYT, there are a lot of models and access is free now. Here's one. They had one that showed the effect of interventions as well, having trouble finding it.

https://www.google.com/amp/s/www.nytimes.com/2020/03/20/health/coronavirus-data-logarithm-chart.amp.html
policywonk98
How long do you want to ignore this user?
AG
I think by end of next week our data will be good enough to take much more precise and targeted actions for most of the US.

I do think NYC and entire west coast will have to essentially be locked down from the rest of the US.

That's my prediction anyway.
Shanked Punt
How long do you want to ignore this user?
A Data problem? I liked his character on Star Trek The Next Generation.


Agnzona
How long do you want to ignore this user?
But none are yet estimating the total already exposed to give us a real death rate.
ac04
How long do you want to ignore this user?
Quote:

Why don't we have good data?
i have heard some version of this question from literally every company i've ever worked for or with (probably 30+ fortune 500 retailers at this point).
law-apt-3g
How long do you want to ignore this user?
Run the numbers on TexAgs posters... exponential math not working.
Agnzona
How long do you want to ignore this user?
Where has spreadsheet Ag been this was his forte!
The Fife
How long do you want to ignore this user?
ac04 said:

Quote:

Why don't we have good data?
i have heard some version of this question from literally every company i've ever worked for or with (probably 30+ fortune 500 retailers at this point).
That's the truth... where I work the unwritten rule is to validate whatever data the system spits out, or look like a complete idiot.
Picard
How long do you want to ignore this user?
AG
Shanked Punt said:

A Data problem? I liked his character on Star Trek The Next Generation.



(cuts him off)

Thank you Mr. Data

Burdizzo
How long do you want to ignore this user?
AG
Agnzona said:

Why don't we have good data? Why can't we run real solid models?

By now we should have a very solid death rate and should be able to extrapolate, deaths based on our actions. i.e. shelter in place, work with social distancing, complete return to normal activity???


I have been asking similar questions about global climate change for the last 20 years.

The answer is "the science is settled" because politicians have already settled it.
Secolobo
How long do you want to ignore this user?
AG
Because politics is always "injected" into the data to serve their intended purpose...
Can I go to sleep Looch?
Joe Exotic
How long do you want to ignore this user?
AG
Rendered Fat said:

It wouldn't matter anyway. Bo Darville, InfectionAg and P.C. Principal are the only ones who understand basic math.


There's a few others, but sadly not many.


We also don't want to kill hundreds of thousands to save your 401k.
Ag87H2O
How long do you want to ignore this user?
AG
Related question. Why do many folks only seem to care about the data regarding the virus and want to ignore the data regarding the economic impact and the resulting negative societal consequences?
Ag87H2O
Agnzona
How long do you want to ignore this user?
I agree and that is certainly harder to quantify, but if the number of suicides per % of unemployment floating around is true we will lose more from that than COVID19 !
IDaggie06
How long do you want to ignore this user?
AG
Quote:

We also don't want to kill hundreds of thousands to save your 401k.

Price worth paying for 350 million to not live in a depression.
Agnzona
How long do you want to ignore this user?
Bo Darville said:

Rendered Fat said:

It wouldn't matter anyway. Bo Darville, InfectionAg and P.C. Principal are the only ones who understand basic math.


There's a few others, but sadly not many.


We also don't want to kill hundreds of thousands to save your 401k.
A better question is why don't you care about the unintended consequence deaths that will occur and why don't you care about all the everyday deaths? Why is this so much more important to you than the other deaths?
Rendered Fat
How long do you want to ignore this user?
Agnzona said:

Bo Darville said:

Rendered Fat said:

It wouldn't matter anyway. Bo Darville, InfectionAg and P.C. Principal are the only ones who understand basic math.


There's a few others, but sadly not many.


We also don't want to kill hundreds of thousands to save your 401k.
A better question is why don't you care about the unintended consequence deaths that will occur and why don't you care about all the everyday deaths? Why is this so much more important to you than the other deaths?


Because he doesn't have a 401k. Very easy to be "noble" when you're broke.
Coog97
How long do you want to ignore this user?
Did the thread that had gotten really contentious get nuked? The one that basically said "the models" showed the virus spreading exponentially, with infections essentially doubling daily and that we'd be approaching 20 million cases world-wide by this weekend... did that exchange disappear?
Joe Exotic
How long do you want to ignore this user?
AG
Agnzona said:

Bo Darville said:

Rendered Fat said:

It wouldn't matter anyway. Bo Darville, InfectionAg and P.C. Principal are the only ones who understand basic math.


There's a few others, but sadly not many.


We also don't want to kill hundreds of thousands to save your 401k.
A better question is why don't you care about the unintended consequence deaths that will occur and why don't you care about all the everyday deaths? Why is this so much more important to you than the other deaths?


Because I've seen no hard numbers showing it will be more. I also don't particularly care about people who commit suicide over money, absent any other mental issues.
k2aggie07
How long do you want to ignore this user?
AG
Agnzona said:

Why don't we have good data? Why can't we run real solid models?

By now we should have a very solid death rate and should be able to extrapolate, deaths based on our actions. i.e. shelter in place, work with social distancing, complete return to normal activity???
I think we have solid data on severity and fatality for observed cases. The challenge is estimating the cases we're not seeing.

There's different ways to try to back into the total using the limited data.

One way is saying, ok for the positive cases we saw, and combining that with what we know (how long between infected and symptoms, how long you're infectious) how many cases should there be out there? That's a spatio-temporal model approach.

Another is saying, let's take a small sample size - like all the expats fleeing from China - and track them. We assume they're a random mix, so they're like a small, randomized study. What does their disease and progression look like? How many of them would not have gone to the doctor but for this event?

Another way is to say - let's assume a fatality rate - how many people would need to have it to produce the deaths we've seen, using the other disease factors? Or, let's assume another parameter, how infectious is it? Then you tweak your model to match up with the fatalities or cases you did see.

Then you can do random testing of a sample group, like that town in Italy... or really broad testing like in S Korea. That may tell you how many people you're not seeing in testing symptomatics only.

The problem is none of these are necessarily representative of what's really going on in a big picture way. So, it's best to accept them all as possible, look at where the general trend is, and each one sort of gives you a range of possibilities. And some that look to be totally contrary to each other may not be - they both may be true, but only true for the sample size they used for their study, their model, whatever.

I think we feel pretty comfortable we know a few things as "the world"
- Observed fatality skews with age, <1%
- Course of the severe disease (including symptoms etc)
- It seems to be very infectious
- There seem to be a very large number of asymptomatic or minor cases
- Social distancing / mass society intervention can reduce case growth. Only telling people wash your hands, stay home, doesn't seem to slow it down very much.


Key question right now has two parts:
- How many are actually asymptomatic?
- How are those spread through the population? By age? By genetics? Are all the <30 y/o the "hidden" carriers the models imply? Or is it a portion of all people?
Agnzona
How long do you want to ignore this user?
Why can't we take the flues estimates % of non diagnosed people and multiple it by the factor (whatever they think that is) of more contagious than flu?

k2aggie07
How long do you want to ignore this user?
AG
We won't really know some of this information until we do some blood / antibody testing on a larger scale. And, maybe this isn't encouraging, but we actually don't even know some of that for seasonal flu. The estimated range of people who are immune for seasonal flu is anywhere between 10% and 60%.

And this is true for a lot of diseases MRSA, strep, ebola, meningococcus .. all have a big part of their transmission done by asymptomatic carriers.
B-1 83
How long do you want to ignore this user?
AG
Agnzona said:

Why don't we have good data? Why can't we run real solid models?

By now we should have a very solid death rate and should be able to extrapolate, deaths based on our actions. i.e. shelter in place, work with social distancing, complete return to normal activity???
Does death rate necessarily correlate with the shear numbers requiring hospitalization and medical assistance? That is where the strain on the system really comes from.
k2aggie07
How long do you want to ignore this user?
AG
You can, but it may not be all that accurate. The CDC estimates that about 50% of people who get the flu, don't go to the doctor, and just suffer at home.

But... there's challenges there. What if this virus can cause a much more minor to much more severe reaction/disease? And if flu is more consistent, that question may or may not tell you a good answer.

Or, what if you get the flu and you know it, bam, in like 2 days you feel like crap. But with this, you may not know, it may be a nagging cough for three weeks? We just don't know.

We've been studying the flu seriously for a hundred years and we still don't know a lot about it. It's absolutely incredible what we've been able to learn and understand about this disease in ~4 months. Really amazing to watch smart people be smart and save lives.
B-1 83
How long do you want to ignore this user?
AG
Agnzona said:

Why don't we have good data? Why can't we run real solid models?

By now we should have a very solid death rate and should be able to extrapolate, deaths based on our actions. i.e. shelter in place, work with social distancing, complete return to normal activity???
Does death rate necessarily correlate with the shear numbers requiring hospitalization and medical assistance? That is where the strain on the system really comes from.
Troutslime
How long do you want to ignore this user?
AG
Bo Darville said:

Rendered Fat said:

It wouldn't matter anyway. Bo Darville, InfectionAg and P.C. Principal are the only ones who understand basic math.


There's a few others, but sadly not many.


We also don't want to kill hundreds of thousands to save your 401k.


How many do you want to kill?
k2aggie07
How long do you want to ignore this user?
AG
Seems to. China the observed fatality (not counting hidden cases) was about 4x higher inside Wuhan than outside of Hubei province.
Agnzona
How long do you want to ignore this user?
ac04 said:

Quote:

Why don't we have good data?
i have heard some version of this question from literally every company i've ever worked for or with (probably 30+ fortune 500 retailers at this point).
So you are saying I have Executive Level potential?
BigOil
How long do you want to ignore this user?
AG
I'm sure there's tons of models... but what data is used to history match and validate the models in order to lend credence to the forecast results?

Typically you would want months and years of data to validate a model, not days and weeks
k2aggie07
How long do you want to ignore this user?
AG
All of the "spread" models I've seen used for this are previously published models for infectious disease, which have parameters. Epidemics all behave in the same way from a mathematic basis. A lot of the studies are repurposing flu pandemic work (both respiratory illnesses).

The new parameters (how long you are infectious, how long incubation, etc) are from observed cases for this disease.

But a lot of the studies being done are backwards looking, like saying - are our assumptions right? That doesn't tell you the real answer, but it can challenge or validate your assumptions.
amercer
How long do you want to ignore this user?
AG
Today's count in Maryland is 424 cases, up from ~400 yesterday. 4 deaths so far.

I don't know when we'll have enough data to really understand the situation, but one side is going to have hell to pay when we do.
Troutslime
How long do you want to ignore this user?
AG
Is there procedural continuity between nations in regards to reporting?

What's the trust factor in regards to other nations and the accuracy of inputs?

And I'm assuming the CDC does have continuity within its reporting system here.
88planoAg
How long do you want to ignore this user?
AG
B-1 83 said:

Agnzona said:

Why don't we have good data? Why can't we run real solid models?

By now we should have a very solid death rate and should be able to extrapolate, deaths based on our actions. i.e. shelter in place, work with social distancing, complete return to normal activity???
Does death rate necessarily correlate with the shear numbers requiring hospitalization and medical assistance? That is where the strain on the system really comes from.
Exactly.

Flipping from forum 16 to the CV board is like night and day.

Hospitals are being overrun. Selective hospitals so far, but expected to increase. Not bodies piling up, but beds being taken. People being sent home with oxygen.
Page 1 of 4
 
×
Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.