General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsWhat am I missing? IMHE (University of Washington) Predicts 72,433 total deaths on August 4th 2020
https://covid19.healthdata.org/united-states-of-america3rd Graph
Total deaths
72,433
COVID-19 deaths projected by August 4, 2020
There are 63,800 now (WorldoMeters.com)
So they predict only 8,600 more deaths in the next 95 days. So an average of 90 more per day!
These people are much smarter than me so I am confused!
Steelrolled
(2,022 posts)So either
1. Their models are not accurate
2. The data is bad
I'd say it mostly is 1.
bullimiami
(13,037 posts)And its constantly changing.
Small changes in the transmission rate yield wild swings in case numbers.
Laura PourMeADrink
(42,770 posts)mitigation tactics. Which I would assume will be " not at all" by Aug. Not with a "leader" who doesn't care about human life. Sadly
ProfessorGAC
(64,413 posts)The data is bad, but death data is fairly concrete. I know they still test, but dead is dead.
So, I think they've got some pretty shaky assumptions as the basis of their models.
The downgraded the estimate 3 times, now pushed it back up twice.
And(!), the number they keep reporting is the bottom of the prediction range. They should be touting the whole range. The upper value of the range for August 1, is 114,000.
So, their model really is 93,000+/-20,000.
Bernardo de La Paz
(48,776 posts)ProfessorGAC
(64,413 posts)No doubt there's cheating on the numbers in some states.
I have no reason to doubt IL, NY, NJ, or CA. Even Ohio seems on the up & up, and Michigan is too awful for suppression. If one was going to lie with the numbers, they would have done better than the bad reports their issuing.
I do like the unexpected death rate comparison. Not as a sole value, but as a good pressure test of other projections.
muriel_volestrangler
(101,149 posts)So that would be "86,800 +/- 27,400" or so. Something in the model says it's more likely to be in the lower part of the range; the 73,000 must be a mean, median or mode of their predictions.
ProfessorGAC
(64,413 posts)Take a look at their model graph. The median line, and bottom of range are zero slope from late May to early August.
But, the upper range still has an upward slope. The middle of the shaded are is 90k.
So they also have an imbalanced +/-. Versus the projection line, the plus is bigger than the minus.
There are some questionable assumptions.
gibraltar72
(7,486 posts)USALiberal
(10,877 posts)gibraltar72
(7,486 posts)over 2,000 again today. I don't see how it's possible. But I'm just a dummy.
USALiberal
(10,877 posts)ProfessorGAC
(64,413 posts)Guy from IHME was on CNN yesterday. They raised it by over 6,000 by August because of that.
Also, I just looked at their graph.
First, their low value for today is already almost 3,000 below actual.
The upper end in August is 114,000. The midpoint is around 93,000.
So, they really seemed to be projecting 30,000 more deaths in May-July.
greenjar_01
(6,477 posts)ProfessorGAC
(64,413 posts)I still find their model suspect, but I am more annoyed by their insistence of publicizing the best case number only.
rickford66
(5,498 posts)Blue_true
(31,261 posts)That would have blown up their model.
the Maryland governor called out the national guard to protect the tests coming in from South Korea because he feared they would be seized by the federal government
uponit7771
(90,225 posts)beachbumbob
(9,263 posts)Buckeye_Democrat
(14,846 posts)It cant be blamed on just bad data or bad assumptions.
If thats IMHEs current projection for August, then they should have the good sense to not publish the nonsense and to keep working on the model in the meantime!
greenjar_01
(6,477 posts)We're damn steady at 2,000+ a day
malaise
(267,797 posts)USALiberal
(10,877 posts)ramblin_dave
(1,546 posts)The default simulation is for maintaining the current interventions for the next 4 months. That shows a death toll of 86,000 by August 31, which seems more realistic.
However, many states are starting to relax interventions. By setting the simulator to show the current interventions for the next 4 weeks, followed by minimal restrictions for 12 weeks, then re-analyzing, a death toll of around 1,100,000 is predicted by August 31.
USALiberal
(10,877 posts)Fiendish Thingy
(15,363 posts)😲😲😲
bullwinkle428
(20,626 posts)have egg on their face at the end of this!
Fiendish Thingy
(15,363 posts)mike_c
(36,213 posts)Up front let me say that I'm not an epidemiologist but I am a scientist who has spent many years modeling nature. Broadly speaking there are two general types of models. Conceptual models depict natural processes and are generally used to test our understanding about them. They are not meant to make accurate predictions about future events. The specific numbers they spit out are rarely accurate, if ever. No one really expects them to be accurate predictors. Rather, they articulate our assumptions and broad understanding of natural processes, such as how populations grow or how diseases propagate within our communities.
The second class of models is statistical. Statistical models are intended to make accurate predictions about future events, however they provide little or no understanding of the biological processes that underlay the data. A basic statistical model says "If the independent variable ranges from x1 to x2, then the dependent variable will likely (we think) range from y1 to y2," and we can know how confident we can be in those predictions. Statistical models are constructed after the fact, i.e. after we acquire sufficient data for analysis. It's also worth noting that statistical models become increasingly shaky when we use them to extrapolate beyond the observed data.
In the present case we still don't know much about this coronavirus' pathogenicity or its epidemiology, so we're trying to make informed decisions based largely on conceptual models of viral transmission and virulence. People need to realize that if such a model predicts X people will die today and Y people will die tomorrow, X+Y coffins is not likely to be the number needed, no matter what the model predicts.
USALiberal
(10,877 posts)MineralMan
(146,190 posts)Squinch
(50,773 posts)record that it's crap.
It ALWAYS drastically under-predicts. That is why Donnie Bodybags loves it.
Crunchy Frog
(26,548 posts)I think they're pretty much worthless.
They seem to have a built in assumption that deaths will precipitously drop to almost nothing 2-3 weeks after the date of their prediction, and then stay that way through August.
D_Master81
(1,822 posts)Theyll say theres some new data and revise it up another 10,000.
Ms. Toad
(33,915 posts)I think they are dreaming.
USALiberal
(10,877 posts)BGBD
(3,282 posts)Is going to kill at least 100k people and that's with the massive stay at home and social distancing we have done.
This rush to "reopen" might push that number up drastically.
We shouldn't have been worried about when we could reopen, but how.
We either needed to fundamentally change the way we approach business and how things are done, or invest heavily in the trype fo infrastructure that we would need to ensure safety for people returning. We did neither.
Blue_true
(31,261 posts)So, at least they are staying consistent. My guess is we hit their new number by this time next week.
mackdaddy
(1,520 posts)First, I have seen discussion where the cases graph has the sharp jump up front, but stays at or near the peak with a slow decline down. They are showing a sharp sudden drop off.
We are now well over 2000 new deaths per day and seem pretty steady at that.
We are only testing seriously ill new patients, and have stalled at only 250k tests per day. Hopefully that will increase.
We have 60k dead for 1 million confirmed cases so about a 6% death rate for confirmed cases.
New cases today were just over 30 thousand. It takes about 3-4 weeks from test to death for the terminal cases.
Assuming the same death rate that is still 6%, then 1800 deaths per day at the end of this month.
At that rate we could easily have 50 to 60 thousand ADDITIONAL deaths just the month of May from already detected confirmed cases.
With all of the Stay Home orders loosening, there is a good chance that the new cases per day will go UP, not down.
I hope I am wrong. Please point out anything I am missing here.
USALiberal
(10,877 posts)USALiberal
(10,877 posts)Brainfodder
(6,423 posts)Otherwise, they should be shamed for that easily surpassed # especially with fools partially opening back up way too soon?
sarcasmo
(23,968 posts)Takket
(21,421 posts)their data/methodology is obviously flawed because their predictions don't even pass the "eyeball" test for any person looknig at a graph. i would consider them to be an unreliable source.
The Magistrate
(95,237 posts)Was the best case out of a wide range, and the whole was based on the assumption everything would be done right, in accordance with the best practices urged by experts in public health. Since there is much being done contrary to such recommendations, and best case projections, as Mr. Murphy has warned us, seldom work out anyway, the thing looks ridiculous, and rightly so.
"Optimists meet bitter disappointments. Pessimists get a pleasant surprise now and then."
muriel_volestrangler
(101,149 posts)and their "best case" is 59,343. The figure they had on April 27th for deaths that had already happened was 55,891; so their best case was just 3,452 more deaths in over 3 months.
(Worldometer's 27th April figure was 56,795; April 28th was 59,265, so 2,470 in one day. There may be a certain amount of smoothing involved that explains the discrepancies, but, by any rational expectation, the best case figure for future deaths was about 2 days in the real world.)
A glance at the 7 day averages for New York state and the rest of the USA suggests a good outcome for the next week would be an average 250 deaths a day in NY, and 1,300 in the rest - over 10,000 in total. https://www.democraticunderground.com/?com=view_post&forum=1002&pid=13375993
That'll be about 76,000. Their model is ridiculous, and they should withdraw it, rather than plugging a few new numbers into it every so often, and acting as if it helps.
The Magistrate
(95,237 posts)Teach me to give people the benefit of the doubt....
Celerity
(42,640 posts)and his legions of MAGAT goons of a massive coverup.
The bottom line is Trump and his allies both in and out of the government are total suppressing the case and death numbers. Fux News is near the top of non-governmental gaslighters on thsi, and have the blood of tens of thousands on their hands, along with Trump himself.
HAMMER that home 24-7, from now until the election.
It simply MUST start happening.
DeminPennswoods
(15,246 posts)for CV19 predictions. https://www.technologyreview.com/2020/03/13/905313/cdc-cmu-forecasts-coronavirus-spread/