Donate to Remove ads

Got a credit card? use our Credit Card & Finance Calculators

Thanks to Anonymous,johnhemming,Anonymous,Rhyd6,dredd0, for Donating to support the site

Coronavirus - modelling aspects only please

Scientific discovery and discussion
GrahamPlatt
2 Lemon pips
Posts: 139
Joined: November 4th, 2016, 9:40 am
Has thanked: 22 times
Been thanked: 30 times

Re: Coronavirus - modelling aspects only please

#310263

Postby GrahamPlatt » May 20th, 2020, 2:20 pm

Nimrod103 wrote:The curious case of all the countries whose death statistics are bogus because of ignoring care home fatalities:

https://order-order.com/2020/05/20/eu-r ... /#comments


Belgium has always included their care home deaths with their overall returns (while UK only reported deaths in hospital)
Germany’s 8000 (current figure) x 37% is some way lower than the UK’s 35000 x 21%
Or to put it another way, UK has 2.5x Germany’s care home deaths.

dealtn
Lemon Quarter
Posts: 1051
Joined: November 21st, 2016, 4:26 pm
Has thanked: 22 times
Been thanked: 354 times

Re: Coronavirus - modelling aspects only please

#310269

Postby dealtn » May 20th, 2020, 2:30 pm

GrahamPlatt wrote:
Nimrod103 wrote:The curious case of all the countries whose death statistics are bogus because of ignoring care home fatalities:

https://order-order.com/2020/05/20/eu-r ... /#comments


a) Where’s the data from? Unattributed.


What makes you think it is unattributed?

Seems to be from The European Centre For Disease Prevention and Control, which Wikipedia says is an independent agency of the European Union.

https://www.ecdc.europa.eu/sites/defaul ... idance.pdf is perhaps a better link to the source, rather than via a political blog.

Best to keep non-science comments to "another place" please as this is the Science Board

feder1
2 Lemon pips
Posts: 212
Joined: November 8th, 2016, 8:28 am
Has thanked: 26 times
Been thanked: 12 times

Re: Coronavirus - modelling aspects only please

#310551

Postby feder1 » May 21st, 2020, 10:26 am

DAK how many cases were reported last night 20 May please?

According to Worldometers this morning 21 may, there is a total of 248293 cases and 35704 deaths.

I seem to have made an error writing down the total cases yesterday morning 20 May and have a negative figure for yesterday's cases - wrong!

swill453
Lemon Quarter
Posts: 3165
Joined: November 4th, 2016, 6:11 pm
Has thanked: 230 times
Been thanked: 952 times

Re: Coronavirus - modelling aspects only please

#310552

Postby swill453 » May 21st, 2020, 10:28 am

feder1 wrote:DAK how many cases were reported last night 20 May please?

According to Worldometers this morning 21 may, there is a total of 248293 cases and 35704 deaths.

I seem to have made an error writing down the total cases yesterday morning 20 May and have a negative figure for yesterday's cases - wrong!

My wife is keeping track of these things too and she made a comment last night that there was an inconsistency. I don't recall the details and she's not around at the moment.

Scott.

dealtn
Lemon Quarter
Posts: 1051
Joined: November 21st, 2016, 4:26 pm
Has thanked: 22 times
Been thanked: 354 times

Re: Coronavirus - modelling aspects only please

#310582

Postby dealtn » May 21st, 2020, 11:23 am

feder1 wrote:DAK how many cases were reported last night 20 May please?

According to Worldometers this morning 21 may, there is a total of 248293 cases and 35704 deaths.

I seem to have made an error writing down the total cases yesterday morning 20 May and have a negative figure for yesterday's cases - wrong!


Remarkably on the BBC Breakfast News this morning (in an interview with London Mayor Sadiq Khan) there were zero new infections in London in the last 24 hours. You would think this would be getting a higher profile in reporting.

swill453
Lemon Quarter
Posts: 3165
Joined: November 4th, 2016, 6:11 pm
Has thanked: 230 times
Been thanked: 952 times

Re: Coronavirus - modelling aspects only please

#310584

Postby swill453 » May 21st, 2020, 11:27 am

swill453 wrote:
feder1 wrote:DAK how many cases were reported last night 20 May please?

According to Worldometers this morning 21 may, there is a total of 248293 cases and 35704 deaths.

I seem to have made an error writing down the total cases yesterday morning 20 May and have a negative figure for yesterday's cases - wrong!

My wife is keeping track of these things too and she made a comment last night that there was an inconsistency. I don't recall the details and she's not around at the moment.

Confirmed, she has a note of 248,293 total cases yesterday and 248,818 the day before.

Scott.

zico
Lemon Quarter
Posts: 1499
Joined: November 4th, 2016, 12:12 pm
Has thanked: 329 times
Been thanked: 439 times

Re: Coronavirus - modelling aspects only please

#310619

Postby zico » May 21st, 2020, 12:35 pm

A mix of data and general musings about new cases.
If it's too long to read, main points are that London likely to have fewer than 10 new infection per day, compared to other regions which have thousands of new infections per day.
Current Government testing may now be finding 40%+ of all new infections.
Estimates show there will still be thousands of new infections daily at the end of May after 8 weeks of lockdown - so we may simply have to live with Covid-19 until a vaccine is discovered. Inevitably, the easing of restrictions will mean the spread of the virus will be higher than it was in the lockdown period.

First off, here's a link to the Cambridge study estimating new cases by region (apparently based on all data from Public Health England plus some modelling assumptions). If you scroll down the graph and look at the regional cases, there's clearly a big regional discrepancy.

https://www.mrc-bsu.cam.ac.uk/now-casting/

Cambridge's estimated new infections by region for May 21st :-
London - 3 (Yes, three!)
East - 524
Midlands 641
North-East & Yorks 2,560 (a whopping 42% of all new infections in England)
North-West 1,180
South-East 594
South-West 399
England 5,960 (this total doesn't quite match the total of the regions because they're based on modelling).
(Cambridge estimate for England new infections on May 28th = 3,960 which would be a 33% reduction).

If these figures are anywhere near correct (and they are probably likely to be roughly correct) then London could already eradicate Covid-19 by closing its borders with the rest of the country. (Of course, it can't realistically do that, but it's certainly at a completely different stage to all other regions).

More generally, given we've had 8 weeks of lockdown and still there are thousands of new cases every day, Covid-19 definitely looks to be here to stay, because it's very unlikely the government will be able to track, test and isolate in sufficient numbers to keep up with the spread of the virus.

Why are we still getting thousands of cases per day? Pure speculation, but if there's a significant percentage of people who ignore the rules, that''ll be enough to keep the virus alive and kicking in the UK.

2nd data source is government's daily figures where testing has found 2,472 positive tests yesterday. Assuming Cambridge estimate of 5,960 new infections in England is correct, UK testing programme is detecting 42% of all new infections. That's pretty good if it proves to be the case.

https://www.gov.uk/guidance/coronavirus ... and-deaths

Govt figures are separated by Pillar 1 (Clinical need cases, NHS and care workers) and Pillar 2 (Other key workers and their families). I'm ignoring Pillar 4 (sampling cases) because positive results aren't provided on those).
Yesterday's cases

Pillar 1
16,287 people tested
35,196 tests carried out
1,048 positive tests (6.4% of people tested were positive, 3% of tests carried out were positive)
Average tests per person = 2.2

Pillar 2
44,457 people tested
118,419 tests carried out
1,424 positive tests (4% of people tested were positive, 2.7% of tests carried out were positive)
Average tests per person = 2.7

Not everyone will be tested more than once, and we would expect a lot of people to be taking their first test. I'm making some assumptions below, because we don't know the breakdown of tests.
Assuming 10,000 pillar 1 tests are having their first ever test (and only tested once), the remaining 6,287 are tested 4 times per day.
Assuming 20,000 pillar 2 tests are having their first ever test (and only tested once), the remaining 24,457 are tested 4 times per day.

For comparison, on 12th April, there were 14,506 tests carried out, with 4,342 positive results (30% of tests were positive).
Cambridge estimate for new infections on 12th April is 65,000 which implies only 7% of new infections were being tested on 12th April.

dealtn
Lemon Quarter
Posts: 1051
Joined: November 21st, 2016, 4:26 pm
Has thanked: 22 times
Been thanked: 354 times

Re: Coronavirus - modelling aspects only please

#310629

Postby dealtn » May 21st, 2020, 12:48 pm

zico wrote:More generally, given we've had 8 weeks of lockdown and still there are thousands of new cases every day, Covid-19 definitely looks to be here to stay, because it's very unlikely the government will be able to track, test and isolate in sufficient numbers to keep up with the spread of the virus.



Well I would say "potentially" not "definitely"

If London can get it down to 0, as reported on BBC today, or 3 the number you have, (without the "track, test and isolate..." you are implying we need) it shows what can be done. It isn't definite, but it is very dependent on "behaviour".

zico
Lemon Quarter
Posts: 1499
Joined: November 4th, 2016, 12:12 pm
Has thanked: 329 times
Been thanked: 439 times

Re: Coronavirus - modelling aspects only please

#310638

Postby zico » May 21st, 2020, 12:56 pm

dealtn wrote:
zico wrote:More generally, given we've had 8 weeks of lockdown and still there are thousands of new cases every day, Covid-19 definitely looks to be here to stay, because it's very unlikely the government will be able to track, test and isolate in sufficient numbers to keep up with the spread of the virus.



Well I would say "potentially" not "definitely"

If London can get it down to 0, as reported on BBC today, or 3 the number you have, (without the "track, test and isolate..." you are implying we need) it shows what can be done. It isn't definite, but it is very dependent on "behaviour".


Yes, if London could do border control around the M25, then it would be Covid-free, but practically, this is unlikely to happen, because of the huge commuting range for London workers.

It's puzzling for me that although London had its peak daily infections at the same time as all other regions, it has dropped much more sharply than all other regions. Possibly some element of London-based herd immunity? Or maybe the huge drop in people using overcrowded public transport is more significant in London? There are probably other plausible explanations but I can't think of any at the moment.

zico
Lemon Quarter
Posts: 1499
Joined: November 4th, 2016, 12:12 pm
Has thanked: 329 times
Been thanked: 439 times

Re: Coronavirus - modelling aspects only please

#310644

Postby zico » May 21st, 2020, 1:01 pm

Uh-oh. If the Cambridge modelling figures are accurate, then the ONS random survey should be showing a reduction compared to its report from last week.

Link to Guardian summary of report.

https://www.theguardian.com/world/2020/ ... e-says-ons

Link to proper ONS report.

https://www.ons.gov.uk/peoplepopulation ... d21may2020

dealtn
Lemon Quarter
Posts: 1051
Joined: November 21st, 2016, 4:26 pm
Has thanked: 22 times
Been thanked: 354 times

Re: Coronavirus - modelling aspects only please

#310646

Postby dealtn » May 21st, 2020, 1:04 pm

zico wrote:
dealtn wrote:
zico wrote:More generally, given we've had 8 weeks of lockdown and still there are thousands of new cases every day, Covid-19 definitely looks to be here to stay, because it's very unlikely the government will be able to track, test and isolate in sufficient numbers to keep up with the spread of the virus.



Well I would say "potentially" not "definitely"

If London can get it down to 0, as reported on BBC today, or 3 the number you have, (without the "track, test and isolate..." you are implying we need) it shows what can be done. It isn't definite, but it is very dependent on "behaviour".


Yes, if London could do border control around the M25, then it would be Covid-free, but practically, this is unlikely to happen, because of the huge commuting range for London workers.



You're missing my point. I am not saying London could do "anything".

You appear to be saying that without "track, test and isolate..." given the number of cases across the UK, Covid-19 will definitely be with us for a long time.

I am merely pointing out that a few weeks ago in London alone you had similar numbers of infections as currently across the UK, and there wasn't the "track, test and isolate..." capability either. Yet now London is very close to being free of new infections.

It can happen, there is nothing definite about it staying. But it very much depends on behaviours. For London to get to that stage much of the time we had a stronger lockdown than we currently have. For the rest of the UK to replicate London's success it might not be as easy, and complacency will threaten it.

Hence why I think "potentially" is a better word than "definitely".

zico
Lemon Quarter
Posts: 1499
Joined: November 4th, 2016, 12:12 pm
Has thanked: 329 times
Been thanked: 439 times

Re: Coronavirus - modelling aspects only please

#310649

Postby zico » May 21st, 2020, 1:12 pm

dealtn wrote:
zico wrote:
dealtn wrote:
Well I would say "potentially" not "definitely"

If London can get it down to 0, as reported on BBC today, or 3 the number you have, (without the "track, test and isolate..." you are implying we need) it shows what can be done. It isn't definite, but it is very dependent on "behaviour".


Yes, if London could do border control around the M25, then it would be Covid-free, but practically, this is unlikely to happen, because of the huge commuting range for London workers.



You're missing my point. I am not saying London could do "anything".

You appear to be saying that without "track, test and isolate..." given the number of cases across the UK, Covid-19 will definitely be with us for a long time.

I am merely pointing out that a few weeks ago in London alone you had similar numbers of infections as currently across the UK, and there wasn't the "track, test and isolate..." capability either. Yet now London is very close to being free of new infections.

It can happen, there is nothing definite about it staying. But it very much depends on behaviours. For London to get to that stage much of the time we had a stronger lockdown than we currently have. For the rest of the UK to replicate London's success it might not be as easy, and complacency will threaten it.

Hence why I think "potentially" is a better word than "definitely".


Yes "potentially" is a better worse than definitely, because the UK could eliminate the virus if we have sufficient will to do it. I just consider it to be extremely unlikely that the UK government and people will in practice keep the lockdown strong enough and for long enough for this to actually happen in practice.

servodude
Lemon Quarter
Posts: 1107
Joined: November 8th, 2016, 5:56 am
Has thanked: 181 times
Been thanked: 238 times

Re: Coronavirus - modelling aspects only please

#310657

Postby servodude » May 21st, 2020, 1:42 pm

dealtn wrote:
feder1 wrote:DAK how many cases were reported last night 20 May please?

According to Worldometers this morning 21 may, there is a total of 248293 cases and 35704 deaths.

I seem to have made an error writing down the total cases yesterday morning 20 May and have a negative figure for yesterday's cases - wrong!


Remarkably on the BBC Breakfast News this morning (in an interview with London Mayor Sadiq Khan) there were zero new infections in London in the last 24 hours. You would think this would be getting a higher profile in reporting.

Holiday for testers? ;)
There have been absolutely NO confirmed cases yet in North Korea yet; which is weird in one way as it borders China, but absolutely not weird in another way.

Just saying
Stay well
-sd

dealtn
Lemon Quarter
Posts: 1051
Joined: November 21st, 2016, 4:26 pm
Has thanked: 22 times
Been thanked: 354 times

Re: Coronavirus - modelling aspects only please

#310662

Postby dealtn » May 21st, 2020, 1:58 pm

servodude wrote:
dealtn wrote:
feder1 wrote:DAK how many cases were reported last night 20 May please?

According to Worldometers this morning 21 may, there is a total of 248293 cases and 35704 deaths.

I seem to have made an error writing down the total cases yesterday morning 20 May and have a negative figure for yesterday's cases - wrong!


Remarkably on the BBC Breakfast News this morning (in an interview with London Mayor Sadiq Khan) there were zero new infections in London in the last 24 hours. You would think this would be getting a higher profile in reporting.

Holiday for testers? ;)
There have been absolutely NO confirmed cases yet in North Korea yet; which is weird in one way as it borders China, but absolutely not weird in another way.

Just saying
Stay well
-sd


Not sure why you are being so dismissive, particularly on this Board. An alternative, above, postulated it was 3.

Whether 0, or 3, that is a remarkable turnaround from the 1,000s previously for a population measured in the millions.

servodude
Lemon Quarter
Posts: 1107
Joined: November 8th, 2016, 5:56 am
Has thanked: 181 times
Been thanked: 238 times

Re: Coronavirus - modelling aspects only please

#310698

Postby servodude » May 21st, 2020, 3:56 pm

dealtn wrote:
servodude wrote:
dealtn wrote:
Remarkably on the BBC Breakfast News this morning (in an interview with London Mayor Sadiq Khan) there were zero new infections in London in the last 24 hours. You would think this would be getting a higher profile in reporting.

Holiday for testers? ;)
There have been absolutely NO confirmed cases yet in North Korea yet; which is weird in one way as it borders China, but absolutely not weird in another way.

Just saying
Stay well
-sd


Not sure why you are being so dismissive, particularly on this Board. An alternative, above, postulated it was 3.

Whether 0, or 3, that is a remarkable turnaround from the 1,000s previously for a population measured in the millions.


Apologies if my scepticism was taken personally; that was certainly not my intent.

You've presented a very good case in previous posts for a bit of dispassion and rigour around the numbers on COVID in the UK
- and I utterly agree with you on that front!

I'd love to see a great big huge whopping turn around in the UK test results - as it increases the chance of me catching up with my family again
- but I'm not going to hang all my hope and want on a single datum
- not when the testing methodologies and reportage are so open to manipulation and the figures spun to the degree that they have been (hence the NK comment)

I'll be the first to cheer and buy a round when this sorts itself out; I'm going to try very hard to not jump up too early and look like a twat (done that often enough to realise it's not the song I thought it was ;) )

- sd

GoSeigen
Lemon Quarter
Posts: 2136
Joined: November 8th, 2016, 11:14 pm
Has thanked: 522 times
Been thanked: 549 times

Re: Coronavirus - modelling aspects only please

#311394

Postby GoSeigen » May 23rd, 2020, 12:50 pm

In #311332 in the Snug:

Mike4 wrote:I'm surprised you cannot see the difference. Heart attacks and ovarian cancers kill people at a broadly steady pace. The number deaths from CV19 left unchecked in a population increases broadly exponentially, until the disease runs out of victims. Or do you not accept that as true?


No, that is not true. The growth of an infections disease approximates the logistic function. This is approximately exponential in the very early stages, but long before the victims have run out the spread slows. Most deaths occur after the exponential stage has ended.

GS

Bubblesofearth
Lemon Slice
Posts: 315
Joined: November 8th, 2016, 7:32 am
Been thanked: 84 times

Re: Coronavirus - modelling aspects only please

#311400

Postby Bubblesofearth » May 23rd, 2020, 1:14 pm

XFool wrote:
It's infectious?

Of course "runs out of victims" can mean many things: Everybody has already been infected, the people still around are mostly immune, a successful vaccination programme. In these cases it presumably ends up being endemic, which it likely will do. But that's a whole new problem.

But why are some people apparently having difficulty with the concept of 'infectious'?



I have no problem with either the concept of infection or that the virus is infectious. I simply asked for evidence for your claim of exponential growth of cases. I've looked at a lot of the data from different countries and fail to see much evidence of exponential growth. Maybe in the very early stages but the data is noisy there.

You could argue that this is because lockdowns were introduced but that doesn't explain Sweden where increases early on were more closely approximated by linear functions than exponentials.

Maybe if no social distancing measures were advised there would have been exponential growth but that is speculation not evidence.

BoE

dealtn
Lemon Quarter
Posts: 1051
Joined: November 21st, 2016, 4:26 pm
Has thanked: 22 times
Been thanked: 354 times

Re: Coronavirus - modelling aspects only please

#311404

Postby dealtn » May 23rd, 2020, 1:19 pm

GoSeigen wrote:In #311332 in the Snug:

Mike4 wrote:I'm surprised you cannot see the difference. Heart attacks and ovarian cancers kill people at a broadly steady pace. The number deaths from CV19 left unchecked in a population increases broadly exponentially, until the disease runs out of victims. Or do you not accept that as true?


No, that is not true. The growth of an infections disease approximates the logistic function. This is approximately exponential in the very early stages, but long before the victims have run out the spread slows. Most deaths occur after the exponential stage has ended.

GS


Not saying you're wrong. I haven't researched it. But I suspect the 2 words "left unchecked" have some bearing on that. Presumably there has been an amount of work leading to the conclusion "the growth of an infections disease approximates the logistic function"? So was that work based on observations in the real world (or to some extent a laboratory), and if so were these on infections that were "left unchecked" (easier in a laboratory than the real world I would think), or on those where attempts were made, either through behavioural change, or medical intervention, to "check" it?

It could be you are not in disagreement with each other, just not realising the subtle, but important, difference in what you are claiming.

SalvorHardin
Lemon Slice
Posts: 861
Joined: November 4th, 2016, 10:32 am
Has thanked: 845 times
Been thanked: 546 times

Re: Coronavirus - modelling aspects only please

#311437

Postby SalvorHardin » May 23rd, 2020, 2:45 pm

America's Center for Disease Control (CDC) has released some new mortality estimates for the Wuhan Virus. A much lower population mortality rate of 0.4%. I found this on my go-to site, Spin Strangeness and Charm.

"Parameter values are based on data received by CDC prior to 4/29/2020. Their R0=2.5 (remember, R-naught is the reproductive number absent any intervention). Percent asymptotic infections is 35%

I'm having problems in displaying the data as a table. Here are the key figures IMHO:

Age under 50. Mortality 0.05%, chance of hospitalisation 1.7%

Age 50 to 64. Mortality 0.2%, chance of hospitalisation 4.5%

Over 65. Mortality 1.3%, chance of hospitalisation 7.4%

Overall. Mortality 0.4%. Chance of hospitalisation 3.4%

CDC link
https://www.cdc.gov/coronavirus/2019-ncov/hcp/planning-scenarios.html#table-1

Spin, Strangeness and Charm link
https://spinstrangenesscharm.wordpress.com/2020/05/23/covid19-update-may-23-2020-cdc-dramatically-revises-fatality-rates-downward-important-new-immunity-data/

The blog estimates that undocumented infections are 22 times the number of official infections. 0.4% mortality is consistent with the larger surveys in Germany and Iceland, in particular when Hendrik Streeck's team tested the town of Gangelt and estimated 0.36% mortality.

It's an old person's disease, which also affects those with certain existing conditions. Young and middle-aged people have little to worry about unless they have one or more of these conditions.

zico
Lemon Quarter
Posts: 1499
Joined: November 4th, 2016, 12:12 pm
Has thanked: 329 times
Been thanked: 439 times

Re: Coronavirus - modelling aspects only please

#311449

Postby zico » May 23rd, 2020, 3:00 pm

Times today reports that excess deaths for Italy during March and April were 47,000 compared to official government figures of 27,938 for the same period. Italy has also been reporting 200 deaths/day during May.

UK figures to date are 36,400 and 55,000 respectively, but these aren't strictly comparable because the UK excess deaths figure only goes up to May 8th so is a couple of weeks behind the daily published figures.


Return to “Science”

Who is online

Users browsing this forum: No registered users and 1 guest