Genuine points of uncertainty

Covid-19 discussion, bring your own statistics
User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: Genuine points of uncertainty

Post by lpm » Tue Jul 13, 2021 1:56 pm

It's a bit of a mystery why hospitalisation numbers are tracking below all the models' forecasts. The gearing can't be too unexpected, the age profile of new cases can't be too unexpected.

But it's going to get a lot worse over the next month.
What ever happened to that Trump guy, you know, the one who was president for a bit?

User avatar
jimbob
Stummy Beige
Posts: 2719
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: Genuine points of uncertainty

Post by jimbob » Tue Jul 13, 2021 5:42 pm

lpm wrote:
Tue Jul 13, 2021 1:56 pm
It's a bit of a mystery why hospitalisation numbers are tracking below all the models' forecasts. The gearing can't be too unexpected, the age profile of new cases can't be too unexpected.

But it's going to get a lot worse over the next month.
The only thing I can think of is that the immunity is increasing as the vaccination is continuing to bite
Have you considered stupidity as an explanation

User avatar
shpalman
Light of Blast
Posts: 4912
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Genuine points of uncertainty

Post by shpalman » Tue Jul 13, 2021 6:31 pm

lpm wrote:
Tue Jul 13, 2021 1:56 pm
It's a bit of a mystery why hospitalisation numbers are tracking below all the models' forecasts. The gearing can't be too unexpected, the age profile of new cases can't be too unexpected.

But it's going to get a lot worse over the next month.
The lag between cases and deaths actually seems to be about three weeks, with a 0.3% CFR, whereas the lag between cases and hospital admissions per day is about two weeks, with about 5% of cases ending up in hospital.

I figured this out by basically shifting the curves around on the semilog plot until they seem to match the recent minimum and then increase; the axes make direct sense only for the cases per day curve.
uk-phases.png
uk-phases.png (62.96 KiB) Viewed 472 times
(It's a three week lag for numbers in hospital and in ventilator beds but that's a different kind of metric, a "currently in this state" rather than "number arriving in this state today", so has a less obvious significance. The factors are 40% and 6%, respectively, whatever that means.)
molto tricky

User avatar
shpalman
Light of Blast
Posts: 4912
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Genuine points of uncertainty

Post by shpalman » Tue Jul 13, 2021 6:40 pm

jimbob wrote:
Tue Jul 13, 2021 5:42 pm
lpm wrote:
Tue Jul 13, 2021 1:56 pm
It's a bit of a mystery why hospitalisation numbers are tracking below all the models' forecasts. The gearing can't be too unexpected, the age profile of new cases can't be too unexpected.

But it's going to get a lot worse over the next month.
The only thing I can think of is that the immunity is increasing as the vaccination is continuing to bite
I've tried to make the point that the current vaccination rate doesn't really achieve much in terms of reducing the susceptible population on a week-to-week basis so it can't really be responsible for what looks like a changing R_t.

But I do remember during the first wave in Italy, that at the beginning of each week it looked like things were peaking but actually it just meant that each week had a slightly slower exponential growth that the previous week (until eventually it did peak of course). I wondered if it was just people reacting to the cases rates they saw on the news and modulating their behaviour.
molto tricky

User avatar
jimbob
Stummy Beige
Posts: 2719
Joined: Mon Nov 11, 2019 4:04 pm
Location: High Peak/Manchester

Re: Genuine points of uncertainty

Post by jimbob » Tue Jul 13, 2021 7:10 pm

shpalman wrote:
Tue Jul 13, 2021 6:40 pm
jimbob wrote:
Tue Jul 13, 2021 5:42 pm
lpm wrote:
Tue Jul 13, 2021 1:56 pm
It's a bit of a mystery why hospitalisation numbers are tracking below all the models' forecasts. The gearing can't be too unexpected, the age profile of new cases can't be too unexpected.

But it's going to get a lot worse over the next month.
The only thing I can think of is that the immunity is increasing as the vaccination is continuing to bite
I've tried to make the point that the current vaccination rate doesn't really achieve much in terms of reducing the susceptible population on a week-to-week basis so it can't really be responsible for what looks like a changing R_t.

But I do remember during the first wave in Italy, that at the beginning of each week it looked like things were peaking but actually it just meant that each week had a slightly slower exponential growth that the previous week (until eventually it did peak of course). I wondered if it was just people reacting to the cases rates they saw on the news and modulating their behaviour.
I don't see how that works in the context of the plot of weekly admissions vs weekly cases. You'd expect on average a decade per decade increase in cases.

Image

just eyeballing the graph, because I CBA to open the data, in the time we've gone from about 20,000 cases per week to 200,000 we've seen hospital admissions go from between 600 & 700 admissions per week to about 3000 so a tenfold increase in cases and just under fivefold increase in admissions.

The same seems to have happened in the second wave, but to a lesser extent.

I probably should use base 2 log scales
Have you considered stupidity as an explanation

Chris Preston
Fuzzable
Posts: 281
Joined: Tue Nov 12, 2019 8:05 am

Re: Genuine points of uncertainty

Post by Chris Preston » Wed Jul 14, 2021 10:08 am

jimbob wrote:
Tue Jul 13, 2021 1:49 pm

This is less severe - by how much it's hard to see, and the "gears" seem to be slipping as a doubling in cases isn't leading to a doubling in admissions. Presumably with vaccine protection improving over time during this wave.
This time around the virus is infecting a younger cohort, who are less likely to need hospital treatment. The data still suggests that if you have been vaccinated and do get infected by the delta variant, you are still protected against hospitalisation.

The current delta variant outbreak in Sydney of 864 cases has put 71 in hospital. Not too different to the first wave rate. That is because less than 30% of the population has received their first vaccine and only 8% are fully vaccinated.

On another note, we currently have no community cases in South Australia, but are likely to get locked down because of three maskless removalists from Sydney and failure of people to check in.
Here grows much rhubarb.


User avatar
basementer
Catbabel
Posts: 925
Joined: Mon Nov 11, 2019 1:03 pm
Location: Aotearoa
Contact:

Re: Genuine points of uncertainty

Post by basementer » Wed Jul 14, 2021 2:15 pm

Footnote about Chris Preston's maskless removalists: when my Australian ex first used the word "removalist" in conversation I thought it sounded like the adherent of a philosophy, like "nihilist". It's someone who works for a removals and storage company.
Money is just a substitute for luck anyway. - Tom Siddell

User avatar
lpm
Stummy Beige
Posts: 3263
Joined: Mon Nov 11, 2019 1:05 pm

Re: Genuine points of uncertainty

Post by lpm » Wed Jul 14, 2021 2:18 pm

If you say "removals men" these days you get arrested.
What ever happened to that Trump guy, you know, the one who was president for a bit?

Herainestold
Dorkwood
Posts: 1159
Joined: Mon Nov 25, 2019 1:23 pm

Re: Genuine points of uncertainty

Post by Herainestold » Wed Jul 14, 2021 5:34 pm

Chris Preston wrote:
Wed Jul 14, 2021 10:08 am
jimbob wrote:
Tue Jul 13, 2021 1:49 pm

This is less severe - by how much it's hard to see, and the "gears" seem to be slipping as a doubling in cases isn't leading to a doubling in admissions. Presumably with vaccine protection improving over time during this wave.
This time around the virus is infecting a younger cohort, who are less likely to need hospital treatment. The data still suggests that if you have been vaccinated and do get infected by the delta variant, you are still protected against hospitalisation.

The current delta variant outbreak in Sydney of 864 cases has put 71 in hospital. Not too different to the first wave rate. That is because less than 30% of the population has received their first vaccine and only 8% are fully vaccinated.

On another note, we currently have no community cases in South Australia, but are likely to get locked down because of three maskless removalists from Sydney and failure of people to check in.
I still think Australia is doing it right, with hard lockdowns and severe penalties for infractions. It gives a space for a comprehensive vaccine roll out.
One shouldnt be too hard on the relevant authorities, who could know beforehand which vccines would work well and what the adverse effects would be? There is still lots of time to do it right and acquire more vaccines.There are still some good ones in the pipeline, like Novavax, and it may have proved advantageous for Australia to wait.
Delta changes everything.

User avatar
Martin Y
After Pie
Posts: 2093
Joined: Mon Nov 11, 2019 1:08 pm

Re: Genuine points of uncertainty

Post by Martin Y » Thu Jul 15, 2021 8:18 am

lpm wrote:
Wed Jul 14, 2021 2:18 pm
If you say "removals men" these days you get arrested.
Though if you say "removalists" you don't get understood. Not around here anyway. It did make me stop and realise I've never heard them called anything but removals men. Removals people seems like it might feasibly catch on.

We've recently been circulated a list of technical terms we've been asked to stop using for inclusiveness (blacklist/whitelist, master clock/slave clock etc) which is all fine except it's not so easy to introduce alternatives that'll catch on. I realise now I can't tell you the recommended substitutes without looking them up again.

tom p
Dorkwood
Posts: 1090
Joined: Mon Nov 11, 2019 1:14 pm
Location: the low countries

Re: Genuine points of uncertainty

Post by tom p » Thu Jul 15, 2021 8:21 am

Martin Y wrote:
Thu Jul 15, 2021 8:18 am
lpm wrote:
Wed Jul 14, 2021 2:18 pm
If you say "removals men" these days you get arrested.
Though if you say "removalists" you don't get understood. Not around here anyway. It did make me stop and realise I've never heard them called anything but removals men. Removals people seems like it might feasibly catch on.

We've recently been circulated a list of technical terms we've been asked to stop using for inclusiveness (blacklist/whitelist, master clock/slave clock etc) which is all fine except it's not so easy to introduce alternatives that'll catch on. I realise now I can't tell you the recommended substitutes without looking them up again.
I've never seen a removals woman. I'm sure they must exist (I remember in Brummagen 25 years ago there was a removals company called Chris Breeze & Wife), but I see a lot of them as people seem to move in and out a lot around here & I have only ever seen blokes lugging the furniture in & out

User avatar
Martin Y
After Pie
Posts: 2093
Joined: Mon Nov 11, 2019 1:08 pm

Re: Genuine points of uncertainty

Post by Martin Y » Thu Jul 15, 2021 8:53 am

It wasn't ever a sexism thing anyway; it was to distinguish them from the chimpanzees.

tom p
Dorkwood
Posts: 1090
Joined: Mon Nov 11, 2019 1:14 pm
Location: the low countries

Re: Genuine points of uncertainty

Post by tom p » Thu Jul 15, 2021 9:39 am

Martin Y wrote:
Thu Jul 15, 2021 8:53 am
It wasn't ever a sexism thing anyway; it was to distinguish them from the chimpanzees.
hahahaha. It'sh the tashte

User avatar
bolo
Catbabel
Posts: 809
Joined: Mon Nov 11, 2019 1:17 pm
Location: Washington DC

Re: Genuine points of uncertainty

Post by bolo » Thu Jul 15, 2021 1:43 pm

They are just called movers here and always have been, so far as I know.

I suppose technically everything they move has been moved at least once previously, so the re- in remove isn't exactly wrong, but it does seem superfluous.

User avatar
bob sterman
Snowbonk
Posts: 580
Joined: Mon Nov 11, 2019 10:25 pm

Re: Genuine points of uncertainty

Post by bob sterman » Thu Jul 15, 2021 3:13 pm

UK COVID ZOE estimated daily new cases of COVID are currently a lot lower than the actual recorded new daily cases of COVID (even by specimen date).

COVID ZOE has daily new cases at about 33,000 - actual confirmed new cases (by specimen date) are > 40,000 for July 12th and 13th - and will climb as late results are added.

Point of uncertainty? How? Will they adjust the ZOE model???

User avatar
sTeamTraen
After Pie
Posts: 1973
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: Genuine points of uncertainty

Post by sTeamTraen » Thu Jul 15, 2021 3:23 pm

Cardinal Fang wrote:
Sun Dec 27, 2020 1:58 pm
Been seeing a lot of noise on social media the last few days about a meta-analysis of anti-parasitic drug ivermectin that supposedly shows it's good for treating COVID

Running my inexperienced eye over it, it seems to me that a lot of the studies seem to be in-vitro, in silico, animal studies or really tiny studies i.e. not very good to begin with studies. But could someone who knows a bit more about these sorts of thing run their eye over. Seems those with a vested interest in proving ivermectin works (who seem to share a significant overlap with those who used to tout HCQ) are lumping anything they can find together, yet it seems to me that one can't compare directly an in-vitro study to a study in hamsters, to a study in humans

https://c19ivermectin.com/

CF
Bump.

I'm not saying that people who are interested in Ivermectin being debunked should keep a close eye on the Guardian over the next 24 hours. But I"m not not saying that either.
Something something hammer something something nail

User avatar
headshot
Catbabel
Posts: 677
Joined: Tue Nov 12, 2019 9:40 am

Re: Genuine points of uncertainty

Post by headshot » Thu Jul 15, 2021 3:57 pm

bob sterman wrote:
Thu Jul 15, 2021 3:13 pm
UK COVID ZOE estimated daily new cases of COVID are currently a lot lower than the actual recorded new daily cases of COVID (even by specimen date).

COVID ZOE has daily new cases at about 33,000 - actual confirmed new cases (by specimen date) are > 40,000 for July 12th and 13th - and will climb as late results are added.

Point of uncertainty? How? Will they adjust the ZOE model???
Probably because fewer people are reporting to the app now. I know I've missed quite a few days recently...just like with my Duolingo too.

User avatar
bob sterman
Snowbonk
Posts: 580
Joined: Mon Nov 11, 2019 10:25 pm

Re: Genuine points of uncertainty

Post by bob sterman » Thu Jul 15, 2021 4:01 pm

headshot wrote:
Thu Jul 15, 2021 3:57 pm
bob sterman wrote:
Thu Jul 15, 2021 3:13 pm
UK COVID ZOE estimated daily new cases of COVID are currently a lot lower than the actual recorded new daily cases of COVID (even by specimen date).

COVID ZOE has daily new cases at about 33,000 - actual confirmed new cases (by specimen date) are > 40,000 for July 12th and 13th - and will climb as late results are added.

Point of uncertainty? How? Will they adjust the ZOE model???
Probably because fewer people are reporting to the app now. I know I've missed quite a few days recently...just like with my Duolingo too.
I've also just realised something I should have considered.

ZOE is modelling infections acquired today right? So given that it normally takes 4-5 days for people to become symptomatic I suppose today's ZOE number should match the number of positive specimens in about 5 days time.

User avatar
shpalman
Light of Blast
Posts: 4912
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Genuine points of uncertainty

Post by shpalman » Thu Jul 15, 2021 6:28 pm

sTeamTraen wrote:
Thu Jul 15, 2021 3:23 pm
Cardinal Fang wrote:
Sun Dec 27, 2020 1:58 pm
Been seeing a lot of noise on social media the last few days about a meta-analysis of anti-parasitic drug ivermectin that supposedly shows it's good for treating COVID

Running my inexperienced eye over it, it seems to me that a lot of the studies seem to be in-vitro, in silico, animal studies or really tiny studies i.e. not very good to begin with studies. But could someone who knows a bit more about these sorts of thing run their eye over. Seems those with a vested interest in proving ivermectin works (who seem to share a significant overlap with those who used to tout HCQ) are lumping anything they can find together, yet it seems to me that one can't compare directly an in-vitro study to a study in hamsters, to a study in humans

https://c19ivermectin.com/

CF
Bump.

I'm not saying that people who are interested in Ivermectin being debunked should keep a close eye on the Guardian over the next 24 hours. But I"m not not saying that either.
https://www.theguardian.com/science/202 ... l-concerns
molto tricky

User avatar
sTeamTraen
After Pie
Posts: 1973
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: Genuine points of uncertainty

Post by sTeamTraen » Thu Jul 15, 2021 7:37 pm

shpalman wrote:
Thu Jul 15, 2021 6:28 pm
sTeamTraen wrote:
Thu Jul 15, 2021 3:23 pm
Bump.

I'm not saying that people who are interested in Ivermectin being debunked should keep a close eye on the Guardian over the next 24 hours. But I"m not not saying that either.
https://www.theguardian.com/science/202 ... l-concerns
Here is my related blog post: https://steamtraen.blogspot.com/2021/07 ... study.html

And the report from the person who brought the paper to my attention: https://grftr.news/why-was-a-major-stud ... retracted/

There's also a Medium piece coming from the redoubtable GidMK (much easier to say than Gideon Meyerowitz-Katz), looking at the implications of this for the meta-analyses of Ivermectin. I'll post that link when he wakes up and sees that he needs to release it. :lol:
Something something hammer something something nail

User avatar
sTeamTraen
After Pie
Posts: 1973
Joined: Mon Nov 11, 2019 4:24 pm
Location: Palma de Mallorca, Spain

Re: Genuine points of uncertainty

Post by sTeamTraen » Thu Jul 15, 2021 8:53 pm

sTeamTraen wrote:
Thu Jul 15, 2021 7:37 pm
shpalman wrote:
Thu Jul 15, 2021 6:28 pm
sTeamTraen wrote:
Thu Jul 15, 2021 3:23 pm
Bump.

I'm not saying that people who are interested in Ivermectin being debunked should keep a close eye on the Guardian over the next 24 hours. But I"m not not saying that either.
https://www.theguardian.com/science/202 ... l-concerns
Here is my related blog post: https://steamtraen.blogspot.com/2021/07 ... study.html

And the report from the person who brought the paper to my attention: https://grftr.news/why-was-a-major-stud ... retracted/

There's also a Medium piece coming from the redoubtable GidMK (much easier to say than Gideon Meyerowitz-Katz), looking at the implications of this for the meta-analyses of Ivermectin. I'll post that link when he wakes up and sees that he needs to release it. :lol:
And here is GidMK's piece: https://gidmk.medium.com/is-ivermectin- ... c079278602
Something something hammer something something nail

User avatar
Woodchopper
Stummy Beige
Posts: 3670
Joined: Sat Oct 12, 2019 9:05 am

Re: Genuine points of uncertainty

Post by Woodchopper » Thu Jul 15, 2021 9:06 pm

Great work, all of you.

User avatar
Cardinal Fang
Fuzzable
Posts: 265
Joined: Tue Nov 12, 2019 7:42 pm

Re: Genuine points of uncertainty

Post by Cardinal Fang » Thu Jul 15, 2021 10:02 pm

sTeamTraen wrote:
Thu Jul 15, 2021 3:23 pm
Cardinal Fang wrote:
Sun Dec 27, 2020 1:58 pm
Been seeing a lot of noise on social media the last few days about a meta-analysis of anti-parasitic drug ivermectin that supposedly shows it's good for treating COVID

Running my inexperienced eye over it, it seems to me that a lot of the studies seem to be in-vitro, in silico, animal studies or really tiny studies i.e. not very good to begin with studies. But could someone who knows a bit more about these sorts of thing run their eye over. Seems those with a vested interest in proving ivermectin works (who seem to share a significant overlap with those who used to tout HCQ) are lumping anything they can find together, yet it seems to me that one can't compare directly an in-vitro study to a study in hamsters, to a study in humans

https://c19ivermectin.com/

CF
Bump.

I'm not saying that people who are interested in Ivermectin being debunked should keep a close eye on the Guardian over the next 24 hours. But I"m not not saying that either.
This would be this article about one of the major studies for ivermectin being withdrawn because of ethical concerns and possible falsifying of data https://www.theguardian.com/science/202 ... l-concerns

CF
Image

User avatar
shpalman
Light of Blast
Posts: 4912
Joined: Mon Nov 11, 2019 12:53 pm
Location: One step beyond

Re: Genuine points of uncertainty

Post by shpalman » Sat Jul 17, 2021 12:58 pm

shpalman wrote:
Tue Jul 13, 2021 6:31 pm
lpm wrote:
Tue Jul 13, 2021 1:56 pm
It's a bit of a mystery why hospitalisation numbers are tracking below all the models' forecasts. The gearing can't be too unexpected, the age profile of new cases can't be too unexpected.

But it's going to get a lot worse over the next month.
The lag between cases and deaths actually seems to be about three weeks, with a 0.3% CFR, whereas the lag between cases and hospital admissions per day is about two weeks, with about 5% of cases ending up in hospital.

I figured this out by basically shifting the curves around on the semilog plot until they seem to match the recent minimum and then increase; the axes make direct sense only for the cases per day curve.

Image

(It's a three week lag for numbers in hospital and in ventilator beds but that's a different kind of metric, a "currently in this state" rather than "number arriving in this state today", so has a less obvious significance. The factors are 40% and 6%, respectively, whatever that means.)
Here's a similar thing done for the second wave last autumn (day 150 was the 19th of July last year):
uk-phases-2ndwave-all.png
uk-phases-2ndwave-all.png (76.05 KiB) Viewed 49 times
I haven't changed the lags; the CFR was 2% (three week lag) and the admissions per day ended up being 12% of the cases per day (two week lag). I've tried to match the upward curve rather than the peak, which may have been influenced by hospital and/or testing capacity saturating, since you're currently on an upward curve and nowhere near the peak in your current wave (which is only going to accelerate since your plan is to remove rather than reintroduce restrictions). (The even less meaningful factors for numbers in hospital and in ventilator beds are 80% and 10% in this graph).

So with your impressive vaccine rollout you've managed to reduce hospital admissions by maybe a bit more than factor of 2. Regarding the saturation of hospital capacity, since the case rate has doubled in the past 15 days then well done you're currently two weeks behind yourselves in the previous wave.

You've also reduced your death rate by not quite an order of magnitude, which buys you about a month extra before oh sh.t it turns out vaccines don't break the link between cases and deaths they just reduce the CFR.

The difference between CFR and hospitalization rate is probably because younger people are ending up in hospital this time. So fewer of them die but they may occupy the bed for longer while they're in there (and be left with long covid).
molto tricky

Post Reply