| An interesting response to the Imperial College pandemic model on 13:28 - Apr 30 with 1704 views | giant_stow |
Cheers for the link. I want the Swedes to be right desperately, but if they are.... |  |
|  |
| An interesting response to the Imperial College pandemic model on 13:34 - Apr 30 with 1683 views | Edmundo |
One of the most sensible, thoroughly considered critiques of this whole sorry saga. If only Matt Handjob would read it... |  |
| |
| An interesting response to the Imperial College pandemic model on 13:43 - Apr 30 with 1673 views | Steve_M |
It looks slightly dubious to me. Saying that the Imperial model hasn't been peer-reviewed and then mentioning antibody studies as a counter, when many of those haven't been peer-reviewed, is a bit inconsistent.
And although science exists by building and critiquing the work of others but this article doesn't really address methodological issues but rather questions the political decisions that such models inform - the costs and benefits of different actions.
India's lockdown was, like many of Modi's policies, seemingly decided on a whim and the consequences of that, particularly on the country's poor, have been detrimental |  |
| |
| An interesting response to the Imperial College pandemic model on 13:53 - Apr 30 with 1646 views | NotSure |
It's encouraging that there are more and more articles like this that are questioning the Imperial College model. I personally don't trust it and I mentioned earlier in the week they refuse to reveal their coding.
Once you've shut down it's incredibly hard to make the decision to open up again. So as Giesecke says, don't shut certain things down in the first place. |  | | |
| An interesting response to the Imperial College pandemic model on 13:55 - Apr 30 with 1635 views | Darth_Koont |
Agree and disagree with a lot of that.
The Imperial College modelling may be overly cautious but it's modelling designed to protect healthcare resources. On that score, it doesn't seem wrong per se - although good that it's going to be updated given there's now a lot more accurate information out there about Covid-19.
But I probably agree with the overall conclusion here. The modelling has been taken literally for the lockdown being the best way to deal with Coronavirus and that it's avoiding 250,000-500,000 deaths. I don't buy the avoidability argument given that we're a year or more away from a vaccine.
It's a fairly narrow take on the spread of the virus and healthcare effects of spikes. That makes it an important study as these need to be limited, of course. But that's against a timeline of a year, of many other fatal illnesses being underreported and missed, of widespread social, economic and mental hardships that will in their turn create significant foreshortening of lives over the next decade.
There are no easy answers but the Imperial College modelling is part of the picture, and I hope we don't regret taking too hard a short-term view because of it. |  |
| |
| An interesting response to the Imperial College pandemic model on 13:57 - Apr 30 with 1632 views | itfcjoe |
| An interesting response to the Imperial College pandemic model on 13:53 - Apr 30 by NotSure |
It's encouraging that there are more and more articles like this that are questioning the Imperial College model. I personally don't trust it and I mentioned earlier in the week they refuse to reveal their coding.
Once you've shut down it's incredibly hard to make the decision to open up again. So as Giesecke says, don't shut certain things down in the first place. |
South Korea were the model to follow 8 weeks ago and are still the model to follow today.
Until we can test, track and trace effectively things won't get back to normal |  |
| |
| An interesting response to the Imperial College pandemic model on 13:59 - Apr 30 with 1619 views | BlueBadger |
| An interesting response to the Imperial College pandemic model on 13:53 - Apr 30 by NotSure |
It's encouraging that there are more and more articles like this that are questioning the Imperial College model. I personally don't trust it and I mentioned earlier in the week they refuse to reveal their coding.
Once you've shut down it's incredibly hard to make the decision to open up again. So as Giesecke says, don't shut certain things down in the first place. |
Considerably harder to justify huge numbers of deaths due to willy-nilly spread an unchecked new disease so that you can have your mates over for a barbecue though. |  |
| |
| An interesting response to the Imperial College pandemic model on 14:11 - Apr 30 with 1573 views | lowhouseblue |
i can't judge the imperial model. they are clearly a very reputable team but this sort of science doesn't give unchallengeable answers.
what i think is interesting is that once imperial had generated it's results the consequences were entirely unstoppable. no government could have stood out against their conclusions - the public, the press, the opposition would have crucified them. criticisms of the methodology wouldn't have been heard. if they had tried to suppress the results it would have been even worse. the simple existence of a report in those terms by that team at that precise timing made it completely irresistible. [Post edited 30 Apr 2020 14:12]
|  |
| And so as the loose-bowelled pigeon of time swoops low over the unsuspecting tourist of destiny, and the flatulent skunk of fate wanders into the air-conditioning system of eternity, I notice it's the end of the show |
| |
Login to get fewer ads
| An interesting response to the Imperial College pandemic model on 14:14 - Apr 30 with 1567 views | StokieBlue |
| An interesting response to the Imperial College pandemic model on 13:53 - Apr 30 by NotSure |
It's encouraging that there are more and more articles like this that are questioning the Imperial College model. I personally don't trust it and I mentioned earlier in the week they refuse to reveal their coding.
Once you've shut down it's incredibly hard to make the decision to open up again. So as Giesecke says, don't shut certain things down in the first place. |
Yes, you did say that and promptly someone posted a link to the code and the datasets.
So perhaps best not to perpetuate that particular falsehood?
SB |  |
| Avatar - IC410 - Tadpoles Nebula |
| |
| An interesting response to the Imperial College pandemic model on 14:15 - Apr 30 with 1560 views | StokieBlue |
| An interesting response to the Imperial College pandemic model on 13:57 - Apr 30 by itfcjoe |
South Korea were the model to follow 8 weeks ago and are still the model to follow today.
Until we can test, track and trace effectively things won't get back to normal |
Zero domestic cases in SK today.
They have done well but we simply weren't setup to implement their model. It also involves a lot of surveillance which some on here would go absolutely crazy about.
SB | |
| Avatar - IC410 - Tadpoles Nebula |
| |
| An interesting response to the Imperial College pandemic model on 14:21 - Apr 30 with 1535 views | itfcjoe |
| An interesting response to the Imperial College pandemic model on 14:15 - Apr 30 by StokieBlue |
Zero domestic cases in SK today.
They have done well but we simply weren't setup to implement their model. It also involves a lot of surveillance which some on here would go absolutely crazy about.
SB |
But isn't that where we are heading with the NHS App? | |
| |
| An interesting response to the Imperial College pandemic model on 14:23 - Apr 30 with 1521 views | StokieBlue |
| An interesting response to the Imperial College pandemic model on 14:21 - Apr 30 by itfcjoe |
But isn't that where we are heading with the NHS App? |
Not to the same level I don't believe. The SK actually made public peoples movements to other people in the area and then people on social media started speculating on who was having what affairs etc and using the movements to justify it.
I assume here the data will remain private and the government will just contact people directly but I've not really looked much into the tracking app.
However, none of that doesn't mean their response wasn't correct. It clearly was as their 0 domestic cases today proves.
SB [Post edited 30 Apr 2020 14:24]
| |
| Avatar - IC410 - Tadpoles Nebula |
| |
| An interesting response to the Imperial College pandemic model on 14:26 - Apr 30 with 1509 views | NotSure |
| An interesting response to the Imperial College pandemic model on 14:14 - Apr 30 by StokieBlue |
Yes, you did say that and promptly someone posted a link to the code and the datasets.
So perhaps best not to perpetuate that particular falsehood?
SB |
Was that a Github link, because according to coders that doesn't have all the necessary information in it. | | | |
| An interesting response to the Imperial College pandemic model on 14:46 - Apr 30 with 1467 views | gordon |
There's quite a lot of misdirection or confused thinking in that article.
The model doesn't 'value' deaths - it just predicts mortality - interpretation or valuing those deaths is the job of politicians.
Antibody testing really hasn't shown anything yet, there have been some pretty flawed attempts so far which have been inconclusive.
Given that in the UK we're probably at about 50,000 covid related deaths now, with penetration probably at about 10-15% (it may be much higher or lower, but this seems a conservative guess), the estimate of 500,000 excess deaths over two years in the absence of intervention in the UK certainly hasn't been disproved, and looks conservative to me.
The other glaring misdirection is the claim that the Imperial Model doesn't look at the negatives of going into lockdown - it's a disease transmission model, not a cost-benefit analysis - so this is akin to complaining that it doesn't take into account the impact on ITFC.
Having said that, it's certainly true that we have very little idea of which measures achieve what, and that the restrictions in Sweden are almost as effective as what we're doing here, with far less of the negative aspects of lockdown. For example, it well may be the case that sitting in the garden with friends bears little risk - we just don't know this one way or another yet.
I posted a link to the model and code the other day, the modelling assumptions / coviates are stated in the paper anyway and it would be relatively simple for modellers to run models appropriate to their own country. Ultimately though, arguing about the models is silly, because in this situation whoever is ultimately proved right by serology, its reasonable to assume that decisions made under massive uncertainty should err on the side of avoiding the risk of thousands of unnecessary deaths.
Additionally, the article almost seems to imply that the results of the Imperial model have been enforced undemocratically on governments, which is daft. |  | | |
| An interesting response to the Imperial College pandemic model on 14:52 - Apr 30 with 1451 views | gordon |
| An interesting response to the Imperial College pandemic model on 14:26 - Apr 30 by NotSure |
Was that a Github link, because according to coders that doesn't have all the necessary information in it. |
I had a look and it looked fine, I don't know anyone who's tried to reproduce the results, but from reading the paper it doesn't look like it would be that hard, so I'd be surprised if people haven't done it.
I'd add that we've basically been comfortably beating the death toll prediction from the 'do nothing' scenario so far, even with interventions - although of course, we don't really know how effective different interventions are. [Post edited 30 Apr 2020 14:53]
| | | |
| An interesting response to the Imperial College pandemic model on 15:02 - Apr 30 with 1417 views | Lord_Lucan |
| An interesting response to the Imperial College pandemic model on 13:57 - Apr 30 by itfcjoe |
South Korea were the model to follow 8 weeks ago and are still the model to follow today.
Until we can test, track and trace effectively things won't get back to normal |
South Korea though have levels of public intrusion tracing and shutting down businesses that no one in the west would tolerate. |  |
| |
| An interesting response to the Imperial College pandemic model on 15:09 - Apr 30 with 1395 views | gordon |
| An interesting response to the Imperial College pandemic model on 13:53 - Apr 30 by NotSure |
It's encouraging that there are more and more articles like this that are questioning the Imperial College model. I personally don't trust it and I mentioned earlier in the week they refuse to reveal their coding.
Once you've shut down it's incredibly hard to make the decision to open up again. So as Giesecke says, don't shut certain things down in the first place. |
I think the first point of 'trusting a model' is false - you shouldn't 'trust' or 'mistrust' a model - models in this context with so much uncertainty around input variables should be used only as one component of decision making.
The second point is reasonable and important - as the lockdown continues and continues, the spring just coils tighter and tighter... | | | |
| An interesting response to the Imperial College pandemic model on 16:34 - Apr 30 with 1314 views | NotSure |
| An interesting response to the Imperial College pandemic model on 14:52 - Apr 30 by gordon |
I had a look and it looked fine, I don't know anyone who's tried to reproduce the results, but from reading the paper it doesn't look like it would be that hard, so I'd be surprised if people haven't done it.
I'd add that we've basically been comfortably beating the death toll prediction from the 'do nothing' scenario so far, even with interventions - although of course, we don't really know how effective different interventions are. [Post edited 30 Apr 2020 14:53]
|
So, after discussing this, the problem was the data for Report 9. The one that showed 2.2 million deaths in the US and 500K deaths here.
The one that caused the government to move from the Swedish model to fuller lockdown. Data in Github apparently only refers to later reports / studies. | | | |
| An interesting response to the Imperial College pandemic model on 16:40 - Apr 30 with 1304 views | gordon |
| An interesting response to the Imperial College pandemic model on 16:34 - Apr 30 by NotSure |
So, after discussing this, the problem was the data for Report 9. The one that showed 2.2 million deaths in the US and 500K deaths here.
The one that caused the government to move from the Swedish model to fuller lockdown. Data in Github apparently only refers to later reports / studies. |
If you think there's an issue with data on the GitHub account, the best thing to do would be to raise it there, their Postdocs respond to queries from other research groups etc. who are looking to use the models.
e.g.
https://github.com/ImperialCollegeLondon/covid19model/issues/91 | | | |
| An interesting response to the Imperial College pandemic model on 16:45 - Apr 30 with 1295 views | NotSure |
| An interesting response to the Imperial College pandemic model on 16:40 - Apr 30 by gordon |
If you think there's an issue with data on the GitHub account, the best thing to do would be to raise it there, their Postdocs respond to queries from other research groups etc. who are looking to use the models.
e.g.
https://github.com/ImperialCollegeLondon/covid19model/issues/91 |
No there is no issue that I know with the data or code on Github.
It is just that they have only released the code for Report 13.
There is nothing for Report 9, which was the big one from Mid March that reported all of the deaths.
It's outdated now but we still seem to be basing our lockdown on it's results.
https://www.imperial.ac.uk/mrc-global-infectious-disease-analysis/covid-19/covid | | | |
| An interesting response to the Imperial College pandemic model on 16:57 - Apr 30 with 1282 views | gordon |
| An interesting response to the Imperial College pandemic model on 16:45 - Apr 30 by NotSure |
No there is no issue that I know with the data or code on Github.
It is just that they have only released the code for Report 13.
There is nothing for Report 9, which was the big one from Mid March that reported all of the deaths.
It's outdated now but we still seem to be basing our lockdown on it's results.
https://www.imperial.ac.uk/mrc-global-infectious-disease-analysis/covid-19/covid |
The model parameters are all stated in the paper, as are the assumptions built into the prediction of 500,000 deaths - anyone with half decent coding skills could rerun those models and get similar numbers. The challenging / controversial / difficult bit is estimating the parameters, and communicating the sensitivity of the results to the uncertainty of those estimates. [Post edited 30 Apr 2020 16:58]
| | | |
| An interesting response to the Imperial College pandemic model on 17:00 - Apr 30 with 1271 views | BloomBlue |
| An interesting response to the Imperial College pandemic model on 13:28 - Apr 30 by giant_stow |
Cheers for the link. I want the Swedes to be right desperately, but if they are.... |
Japan is the interesting model. They haven't locked-down it seems because 1) it's against their constitution and 2) the Countries near them South Korea, Hong Kong & Taiwan never locked-down and had very low death totals whereas as the EU countries/US who have locked-down have high death totals
Plus Japan aren't testing they're only testing if the person as pneumonia and since Feb they've only tested 11,000.
Throw in Japan is densely populated Tokyo has 9million people and in theory they should have very high death numbers.
If they come out of this with really low totals, that will throw out all the theories |  | | |
| An interesting response to the Imperial College pandemic model on 17:05 - Apr 30 with 1263 views | StokieBlue |
| An interesting response to the Imperial College pandemic model on 17:00 - Apr 30 by BloomBlue |
Japan is the interesting model. They haven't locked-down it seems because 1) it's against their constitution and 2) the Countries near them South Korea, Hong Kong & Taiwan never locked-down and had very low death totals whereas as the EU countries/US who have locked-down have high death totals
Plus Japan aren't testing they're only testing if the person as pneumonia and since Feb they've only tested 11,000.
Throw in Japan is densely populated Tokyo has 9million people and in theory they should have very high death numbers.
If they come out of this with really low totals, that will throw out all the theories |
https://metro.co.uk/2020/04/30/japanese-island-hit-second-coronavirus-wave-lockd
SB | |
| Avatar - IC410 - Tadpoles Nebula |
| |
| An interesting response to the Imperial College pandemic model on 18:31 - Apr 30 with 1209 views | BloomBlue |
Every Country will have deaths from CV, it's the total at the end of this that will count and if Japan's approach results in a very low total in what s a very densely populated country and the same for Hong Kong, South Korea than the experts will have to weigh up if lockdown Europe/US style versus Japan (and other Asia Countries) was the ultimate correct approach | | | |
| |
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
