Posts: 3,212
Threads: 103
Joined: May 2009
kander - "man forgets parachute while skydiving, dies of covid-19. he was 32 years old" types of problems in the data.
Right... I'm sure that's the problem with the model. I think I'll stick with the insights of people who actually work with data and modelling.
https://www.statnews.com/2020/04/17/infl...itics-say/
"IHME uses neither a SEIR nor an agent-based approach. It doesn’t even try to model the transmission of disease, or the incubation period, or other features of Covid-19, as SEIR and agent-based models at Imperial College London and others do. It doesn’t try to account for how many infected people interact with how many others, how many additional cases each earlier case causes, or other facts of disease transmission that have been the foundation of epidemiology models for decades.
Instead, IHME starts with data from cities where Covid-19 struck before it hit the U.S., first Wuhan and now 19 cities in Italy and Spain. It then produces a graph showing the number of deaths rising and falling as the epidemic exploded and then dissipated in those cities, resulting in a bell curve. Then (to oversimplify somewhat) it finds where U.S. data fits on that curve. The death curves in cities outside the U.S. are assumed to describe the U.S., too, with no attempt to judge whether countermeasures —lockdowns and other social-distancing strategies — in the U.S. are and will be as effective as elsewhere, especially Wuhan."
Much more at link for those inclined towards reading before opining...
Posts: 11,019
Threads: 750
Joined: Sep 2012
so many deaths are being attributed to wuflu when there was an underlying morbidity.
Wouldn't the deceased need a positive COVID-19 test result, for that to be listed as the cause of death by a doctor? Given that testing is still inadequate, isn't it far more likely that a deceased person might die of COVID-19 without being tested, and something else is listed on their death certificate?
"I'm at that stage in life where I stay out of discussions. Even if you say 1+1=5, you're right - have fun." - Keanu Reeves
Posts: 14,115
Threads: 424
Joined: Aug 2012
https://www.staradvertiser.com/staradver...pollresult
D. No, not at all (485 Votes)
B. Yes, at clinic/doctor’s office (22 Votes)
C. No, tried but didn’t meet criteria (19 Votes)
A. Yes, via drive-thru testing (15 Votes)
Posts: 1,446
Threads: 12
Joined: Oct 2016
quote:
Originally posted by ironyak
kander - "man forgets parachute while skydiving, dies of covid-19. he was 32 years old" types of problems in the data.
Right... I'm sure that's the problem with the model. I think I'll stick with the insights of people who actually work with data and modelling.
https://www.statnews.com/2020/04/17/infl...itics-say/
"IHME uses neither a SEIR nor an agent-based approach. It doesn’t even try to model the transmission of disease, or the incubation period, or other features of Covid-19, as SEIR and agent-based models at Imperial College London and others do. It doesn’t try to account for how many infected people interact with how many others, how many additional cases each earlier case causes, or other facts of disease transmission that have been the foundation of epidemiology models for decades.
Instead, IHME starts with data from cities where Covid-19 struck before it hit the U.S., first Wuhan and now 19 cities in Italy and Spain. It then produces a graph showing the number of deaths rising and falling as the epidemic exploded and then dissipated in those cities, resulting in a bell curve. Then (to oversimplify somewhat) it finds where U.S. data fits on that curve. The death curves in cities outside the U.S. are assumed to describe the U.S., too, with no attempt to judge whether countermeasures —lockdowns and other social-distancing strategies — in the U.S. are and will be as effective as elsewhere, especially Wuhan."
Much more at link for those inclined towards reading before opining...
Well, maybe I'm reading it wrong but it appears that after finding fault with model for most of the article the last two paragraphs recognize that 2 other models have recently been agreeing with IHME.(?)
Posts: 3,212
Threads: 103
Joined: May 2009
Yeah, I would say that is a misreading, unless you're just saying that an increasing number of models are predicting less cases and mortality than before (which is what happens when you put the more recent data into models as it accounts for all the social distancing that is happening) The rest of the predictions for the IHME (peak time and height, plateauing, curve duration, have all not matched real world data very well (as all the model does is take early data such as from Wuhan and Italy, mirror the early portion of the data to create an artificial symmetry on the downward slope, and then try to match everyone's current numbers to those curves)
Basically the effectiveness of social distancing and people's adherence to it has had more of an effect than even the more conservative of estimates given by the models.
The good news - social distancing is working.
The bad news - Downward trends do not mean the virus has gone away or we've reached herd immunity (need 50-66% infections for this to be true - estimates for the US are around 3% in some areas)
The big question - how much can the restrictions be lifted before restarting the exponential curve of new cases and deaths? Time will tell, and the models have next to no data to go by in making such predictions.
The wild card - how much longer will people choose to participate? Many states already have protest groups saying how it is all a hoax and being goaded on by the President to break stay-at-home requirements. Their curves for new cases will be telling in the coming weeks to see the potential costs in human health and deaths for such "expressions of freedom".
Posts: 1,446
Threads: 12
Joined: Oct 2016
quote:
Originally posted by ironyak
Yeah, I would say that is a misreading, unless you're just saying that an increasing number of models are predicting less cases and mortality than before..
Right, because those are the two most important concerns IMO.
Peak time and height have been fairly accurate too with IHME.
Of course, this can (and likely will) be all thrown out the window if restrictions in place are lifted prior to the end of May.
Beyond that, no one knows. Entirely unlikely virus just disappears as suggested by their graphs.
Posts: 3,212
Threads: 103
Joined: May 2009
Durian Fiend - Peak time and height have been fairly accurate too with IHME.
Um no, they haven't. If you want a full criticism of the IHME model, see this from a week ago. (I think you may just be comparing the most recent revision to the near term results, but if you look across the IHME predictions they have been wildly off, but are used to guide policy at the highest levels)
https://www.nationalreview.com/corner/co...nreliable/
"As I detailed in the last post, the revised April 5 model was grossly wrong even in predicting conditions that would obtain on April 5 itself. It had predicted that on that day, New York, the epicenter of the crisis, would need about 24,000 hospital beds, including 6,000 ICU beds. In fact, the model was off by a third — New York had 16,479 hospitalized COVID patients, 4,376 that were in ICU.
On April 8, IHME reduced the total number of hospital beds it had predicted would be needed nationally by a remarkable 166,890 — down to 95,202 from the 262,092 it had predicted less than a week earlier (i.e., it was nearly two-thirds off). The ICU projection over that same week was cut in half: to 19,816 on April 8, down from 39,727 on April 2. The projected need for ventilators also fell by nearly half, to 16,845 from 31,782."
more at link
Posts: 1,446
Threads: 12
Joined: Oct 2016
The "beds needed" numbers have always had an enormous range, from very small to very large. Definitely not useful for predicting, IMO.
Nate Silver's twitter account is always an interesting read. Regarding IHME, he says:
"There are some good critiques of the IHME model in here IMO but this line bothered me. Model outputs *should* change in response to new data and when you're dealing with nonlinear systems, small changes can have a reasonably large effect." https://statnews.com/2020/04/17/influent...itics-say/
He goes on to suggest critics should build their own model, which in fact has been done here: https://covid-19.tacc.utexas.edu/projections
Posts: 1,099
Threads: 71
Joined: Jan 2009
“That the IHME model keeps changing is evidence of its lack of reliability as a predictive tool,”
Listen carefully. Models should not change due to real world data or they lack reliability.
Cheers,
Kirt
Posts: 1,099
Threads: 71
Joined: Jan 2009
[quote] Originally posted by ironyak
https://www.nationalreview.com/corner/co...nreliable/
[/i]
Ok. Now your doing national review. From the article...
“The question is one of balance. American lives are being shattered by the restrictions that have been put in place. The decision to do that was based on models. Those models have no credibility. They now tell us that about 61,000 may die of coronavirus this year — although, if the last few days are any indication, that number could be revised downward soon, perhaps substantially.”
I know I shouldn’t directly respond to your posts (as I probably do to avoid feeding the troll— oops, did I just fail? I think so) but, well, there it is. Too many notes...
Cheers,
Kirt
|
