Author Topic: COVID-19 Data Model Updated Daily  (Read 20194 times)

Area 10

  • Cortez Bank Status
  • *****
  • Posts: 3976
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #180 on: April 06, 2020, 03:43:08 PM »
The stats about infections or death rates in Europe are largely nonsense. People with very obvious COVID symptoms are not being tested in the UK and elsewhere. The hospitals are often actively avoiding testing.

So as far as modelling goes, garbage in, garbage out.

IMO watch instead the total number of deaths in the country and how that differs from seasonally-adjusted norms.

But Staying at home has led to a decrease in deaths due to cardiac problems, road traffic accident, head injury etc. So even that data is hard to interpret.

We donít really know what is going on, Iím afraid, in Europe. What we are discovering is that this disease can cause heart inflammation and also brain inflammation in some people and that this might be causing a significant number of the deaths rather than respiratory problems per se. So the recordings of deaths when testing is not being done is error-prone: what is recorded?

We are learning about this thing as we go along, and mostly the data is of such poor quality that itís hard to draw any conclusions that are more than vaguely guided speculation. Maybe you have better data in the US.


pdxmike

  • Cortez Bank Status
  • *****
  • Posts: 6184
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #181 on: April 06, 2020, 09:54:26 PM »
With the reduction in travel and general @ risk behavior the Net number may be negative.
I can't help but think vehicular deaths are going to be down significantly this year.

On the other hand, I'd assume there will be some additional deaths due to people not going in for or delaying non-virus medical things--tests, preventative stuff...




Admin

  • Administrator
  • Cortez Bank Status
  • *****
  • Posts: 5258
    • View Profile
    • StandUpZone
    • Email
Re: COVID-19 Data Model Updated Daily
« Reply #182 on: April 07, 2020, 02:24:12 AM »
The stats about infections or death rates in Europe are largely nonsense. People with very obvious COVID symptoms are not being tested in the UK and elsewhere. The hospitals are often actively avoiding testing.

So as far as modelling goes, garbage in, garbage out.

IMO watch instead the total number of deaths in the country and how that differs from seasonally-adjusted norms.

But Staying at home has led to a decrease in deaths due to cardiac problems, road traffic accident, head injury etc. So even that data is hard to interpret.

We donít really know what is going on, Iím afraid, in Europe. What we are discovering is that this disease can cause heart inflammation and also brain inflammation in some people and that this might be causing a significant number of the deaths rather than respiratory problems per se. So the recordings of deaths when testing is not being done is error-prone: what is recorded?

We are learning about this thing as we go along, and mostly the data is of such poor quality that itís hard to draw any conclusions that are more than vaguely guided speculation. Maybe you have better data in the US.

Hi Area10,

You are right.  Check out this page: https://ourworldindata.org/coronavirus-testing-source-data# .  They intended to follow testing over time in each country.  It looks like they eventually found it too problematic to continue.  They go over the issues in each country.

I learned that in the USA in many states negative tests are not being reported at all so while we can get a raw number of positives we cannot get a number of positives/people tested. That would be super relevant as it could tell how representative our testing results have been.  This is particularly vexing because they have this information.  The system was just not in place for it to be entered.  That is aside from the other issues caused by insufficient test availability, etc.  Same is true here of naming cause of deaths in ambiguous cases (deaths where testing has not occurred).  I read that we are following no one standard on how to do this from state to state.  We don't know the level of false attribution.  I red two articles about this yesterday which came to very different conclusions.

On the upside, it does seem possible that there will be sufficient motivation going forward to improve collection systems and coordinate testing information globally.  One the one hand we are witnessing the first near real time sharing of international data in the moment of a crisis.  That makes analysis like Beasho is doing possible.  That is pretty remarkable if you think about it.  On the other hand it is frustrating to everyone to have that data be so problematic.  Granted, nothing ever goes smoothly in the moment of a crisis but it is so much worse when the systems are not in place.

« Last Edit: April 07, 2020, 03:12:25 AM by Admin »

ninja tuna

  • Peahi Status
  • *****
  • Posts: 520
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #183 on: April 07, 2020, 04:44:07 AM »
admin,

is this what you are looking for in terms of testing vs cases

https://www.worldometers.info/coronavirus/country/us/

Admin

  • Administrator
  • Cortez Bank Status
  • *****
  • Posts: 5258
    • View Profile
    • StandUpZone
    • Email
Re: COVID-19 Data Model Updated Daily
« Reply #184 on: April 07, 2020, 07:30:24 AM »
Hi Ninja,

You are absolutely correct that a lot of testing data has been collected.   The site that you linked is showing the testing data that is in the system.  My understanding of what the ourworldindata site was trying to do was to justify that data in a meaningful way (make sense of the inconsistencies that they were finding in each country).  As far as I can see they have stopped doing that. 

LaPerouseBay

  • Teahupoo Status
  • ******
  • Posts: 1890
  • downwind dilettante
    • View Profile
    • Email
Re: COVID-19 Data Model Updated Daily
« Reply #185 on: April 07, 2020, 11:58:59 AM »
Support your local shaper

LaPerouseBay

  • Teahupoo Status
  • ******
  • Posts: 1890
  • downwind dilettante
    • View Profile
    • Email
Support your local shaper

SUP Leave

  • Sunset Status
  • ****
  • Posts: 477
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #187 on: April 08, 2020, 08:14:01 AM »
The IMHE (UW) (CDC) model moved down again today to 60k deaths by August (from 82k on Monday). This model had already taken into account the effect of social distancing, but I am not sure to what degree. I know early in the outbreak they assumed 50% social distancing. So I am guessing that the current model must be starting to use some of the effective treatment rates.

So, Beasho's 50k deaths is still probably a good number for this week, but 30k is looking more accurate.

Now I am waiting to see a plan for re-opening the country. I hope each state is working hard on this. The plan cannot be "everything is back to normal" infections would spike, but I also don't think it can be you have to have an antibody license to move around. Guessing it will be some form of relaxed social distancing, but they should definitely make outdoor activities (work and play) start back up first. These are easy to social distance and healthy.
Make paddleboarding great again!

Admin

  • Administrator
  • Cortez Bank Status
  • *****
  • Posts: 5258
    • View Profile
    • StandUpZone
    • Email
Re: COVID-19 Data Model Updated Daily
« Reply #188 on: April 08, 2020, 09:01:02 AM »
Yesterday was an awful day in terms of deaths.  A portion of many models is being based on the death data from  other countries.  Trying to align curves to adjust for start times and use their curves to gauge what our own might be.  There was a false top a few days ago that gave hope that we were later in the curve than had been thought.  Watching this unfold you can see the real world problems in trying to predict this stuff.

« Last Edit: April 08, 2020, 09:53:55 AM by Admin »

Tom

  • Cortez Bank Status
  • *****
  • Posts: 2892
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #189 on: April 09, 2020, 12:11:17 PM »
Here's a good interactive graph that shows projections for countries, and states

https://covid19.healthdata.org/united-states-of-america

Admin

  • Administrator
  • Cortez Bank Status
  • *****
  • Posts: 5258
    • View Profile
    • StandUpZone
    • Email
Re: COVID-19 Data Model Updated Daily
« Reply #190 on: April 09, 2020, 12:48:31 PM »
That is a really nice set of graphs for the IMHE model.  Here is an excellent article on that model (and others):

https://freebeacon.com/coronavirus/a-popular-covid-19-model-just-got-rosier-dont-read-too-much-into-it/

A bigger issue, however, is that the IHME model makes very strong assumptions about how effective and widespread social distancing will be. It assumes that just 3 percent of the population will end up infected by SARS-CoV-2, compared with ranges of 25 to 70 percent in other models. Such an outcome is only possible if social distancing measures are comprehensive and remain in place through the end of May.

TallDude

  • Cortez Bank Status
  • *****
  • Posts: 5182
  • Capistrano Beach
    • View Profile
    • Email
Re: COVID-19 Data Model Updated Daily
« Reply #191 on: April 09, 2020, 01:03:30 PM »
Such an outcome is only possible if social distancing measures are comprehensive and remain in place through the end of May.
Until when....?
It's not overhead to me!
8'8" L-41 ST and a whole pile of boards I don't use.

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2579
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #192 on: April 10, 2020, 06:08:36 AM »
I have been screaming about CoMorbidity since I saw the statistics 3 weeks ago out of China then confirmed by Bloomberg in Italy and the data from Imperial College.  All 100% aligned on the risk to PRIOR HEALTH CONDITIONS.

Then on Sunday 60 Minutes was reporting and the doctor said "We see a high correlation with Obesity."  BMI > than 40 is listed as Morbid Obesity.  And there it was the word again.

Obesity.  Might the Corona virus just target Obesity. 

Obesity as the Leading indicator of death NOT as the Last of a short list of CoMorbidities.

Obesity, Heart Disease, Cancer, Respiratory Disease . . .


PS: The President knows this.  He was at a press conference 3 weeks ago citing how terrible the disease is.  He said something like "I have a friend with the virus.  One day he is fine 3 days later he is in a coma on a ventilator.  It's vicious . . He was a little heavy . . but its terrrible."  He knows.

---------
Reuters on Obesity:
ďThis virus is terrible, it can hit young people, in particular obese young people. Those who are overweight really need to be careful,Ē Delfraissy told France info radio.
https://www.reuters.com/article/us-health-coronavirus-france-confinement/obesity-is-major-covid-19-risk-factor-says-french-chief-epidemiologist-idUSKBN21Q0S7

World Obesity at Risk with BMI >> 25
 ďGiven the extremely high rates of obesity around the globe we expect that a high percentage of the population who will contract coronavirus will also have a BMI over 25.Ē
https://www.worldobesity.org/news/statement-coronavirus-covid-19-obesity

Doctors on Obesity (in Connecticut):
Dr. John Morton who heads up Bariatric Surgery with Yale Medicine explains, ďThereís something called hypoventilation syndrome and that is when you have extra tissue around your chest.  Itís harder to take deep breaths.Ē
Obesity also increases the risk for health issues such as diabetes and high blood pressure.
https://www.wtnh.com/news/health/coronavirus/what-you-need-to-know-about-obesity-and-covid-19/

ZERO NET DEATHS Theory:
« Last Edit: April 10, 2020, 06:23:29 AM by Beasho »

Rider

  • Rincon Status
  • ***
  • Posts: 210
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #193 on: April 10, 2020, 07:25:24 AM »
Is Stage-4 TDS considered an underlying morbidity?

Beasho

  • Cortez Bank Status
  • *****
  • Posts: 2579
    • View Profile
Re: COVID-19 Data Model Updated Daily
« Reply #194 on: April 10, 2020, 07:26:06 AM »
Live model here. 

I modified my 30 day death estimate to use exponential smoothing.  In other words you can't go EXPONENTIAL forever.  Brings down the 30 day estimate to 46K. 

New State Heatmap shown here.  What states are performing WELL as measure by Days to Double and Which are NOT performing well. 

NY, LA are emerging from the high burn phase.  Although NY now has more cases than any other SINGLE country (~ same as Spain).

https://app.powerbi.com/view?r=eyJrIjoiNzExYmM2ZWQtZWVmMS00OTI1LWE1NzYtMTU2OTk3ZGE4ZjMyIiwidCI6IjQzODE2ODE2LTY0OTYtNDU4My1iMmVmLWU5Y2U3MTg1NjI4MCIsImMiOjF9