UKC

As the original thread's got too big we'll carry on chewing the fat here.

In response to SidH

In the past when levels were high, possibly.  They now have established community seeding now though, in the past week they identified 49 traveller cases, 234 'seeking healthcare', 4 hospital admissions testing,  34 planned workforce testing, 5 'cohort screening' and 176 contact traced (which may or may not include fellow travellers as they're testing everyone on arrival anyway). Their 'contacts per case' has steadily declined from mid-teens in June to about 8 today (I've just crunched the stats).

I was going to save it for Saturday, but seeing as we have a continuation…

Awareness of the plan is certainly shaping up, more formal recognition in messaging, it’s being picked up internationally, and the chaos around isolation orders is rising.

> International expert critisism of the government's approach.

My main criticisms are around further reopening when that runs a risk for healthcare, lack of proper explanation of the apparent plan up front, and the ass backwards combination of dropping some of the final restrictions whilst preserving isolation orders.

A fair bit of the international criticism seems to be about allowing an exit wave to happen.  That’s a more fundamental issue than the characteristically shambolic details and messaging.  As fundamental issues go, assuming the slow down in first doses is down to demand not supply (and I would really appreciate solid clarification here, it’s a critical point IMO), other than waiting for second doses to complete the only real options seem to be an exit wave or endless restrictions.  

Where would we be without A/B testing of policy ey?

Funny how VSisjustascramble has gone silent after doggedly criticising me for risking a worse exit wave by advocating caution over healthcare now. I’m getting deja vu - again - as news stories appear of hospitals cancelling significant chunks of non covid business start coming out.

More concerning than admissions are the trends in ITU occupancy doubling times; the last few days of data for England suggest circa 11 days there too.

There’s still a lot of headroom - although the people behind it must be exhausted by now to say the least so I don’t know how much sooner they headroom will become hard to realise, even before considering isolation orders.  Naively speaking it seems odd to me that double vaccinated staff exposed to covid are required to stay home without symptoms and not work on a covid ward where all staff have risk control measures intended to be effective against highly symptomatic patients at a time the plan is to have a massive wave of covid.  I can see why unpicking the “correct” way to navigate that obstacle course is difficult, which is why I’d rather it had been consulted on and done some time ago.

But with big uncertainty over if and when herd immunity thresholds are going to kick in, we need to keep data driven caution as we eat in to that headroom.  Perhaps 35 days at current doubling times before it’s gone.  We’ve got the best chance yet of not overloading it and having the wave in cases break first, but given the uncertainties I still think caution is the critical factor.

I think the government assumed England wouldn't get very far in the Euros and thus wouldn't have all the super-spreading footy watching events going on, last night and the coming final day will have a noticeable effect on infection rates I'm sure. Which is a shame, because my stats show the increase in live case rate had peaked again.

I'm not so sure about that view on international critisism. I see it more as pointing out the science is saying our infection levels are dangerously high than arguing against an exit strategy per say. I like you supported the idea of an exit strategy, based on data, this summer but the government with poor messaging, leaky borders, a failure to improve well known failings in track and trace, some mad decisions, constant lies and defending rank hypocrisy seem to have potentially buggered things up despite the success in vaccination backed by the related good news in ONS antibody data.

The future modeling isn't showing crystal clear success in this attempt at exit, the view is more, as you might say, turdly !

The bad consequences like variant generation and health consequences come down to the total number of infections, and that's a separate issue and on that's more fundamentally tied to the exit wave and the level of restrictions we are aiming for after the wave than the infection levels, which are a function of the total number and how slow or fast we want to get through them.  There's conflation there in some of the reporting IMO.  The main concern with levels are that too slow risks a worse exit wave according to modelling (which we must remember is modelling and not a time machine in to the future) and too high risks healthcare overload; those define the danger in levels to me - somewhat intangibly so.   

Bordering on stummy beige.

Sure there is conflation but some of that is in the chinese whisper and spin of the press reporting. If you listen to what the expert critics actually say they nearly always seem to be a lot clearer in their messaging. Some restrictions being maintained to autumn could mean a large range of options, depending on the data at the time, from just restrictions with little downside like mask use indoors in public to maybe a few more restrictions than now if numbers get very high (its got to be a coin toss on forced extra restrictions at some point soon if case numbers don't slow by decision time next week and large proportion of hospitals have to cancel all non urgent procedures).

https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(21)01589-0/...

I  agree with much of that Lancet opinion piece (mostly from India SAGE members), but ...

In light of these grave risks, and given that vaccination offers the prospect of quickly reaching the same goal of population immunity without incurring them

With regards "same goal", they are not the same, the immune responses are different - broader epitope, some perhaps with less variant sensitivity, from natural infection, a more consistent dose and so response with vaccination (potentially stronger than mild natural infection?).

With regards "quickly", at the UK level, first does are down to < 100,000 per day now and there's about 7.7 million adults currently without a first dose.

• At the current rate, it's not "quickly" it's closer to 3 months which takes us in to the autumn.  

The critical question is why is the rate now so low - supply or demand?

• If it's supply this is a scandal - especially set in the context of current rising rates let a long a reduction in control measures soon and needs addressing urgently
• If it's demand, their argument falls apart.]

Why isn't the media running with the falling vaccination rates and pressing cabinet for information?  This point is absolutely critical to everything.  It's easy to assume it's down to falling demand but it must be clarified IMO.

Their alternative proposal is to below (my emphasis) and I have two problems with it - although I entirely support the call to keep going with current control measures and border control, and to further invest in ventilation .

Instead, the government should delay complete re-opening until everyone, including adolescents, have been offered vaccination and uptake is high, and until mitigation measures, especially adequate ventilation (through investment in CO2 monitors and air filtration devices) and spacing (eg, by reducing class sizes), are in place in schools. Until then, public health measures must include those called for by WHO (universal mask wearing in indoor spaces, even for those vaccinated), the Scientific Advisory Group for Emergencies (SAGE), the US Centers for Disease Control and Prevention (ventilation and air filtration), and Independent SAGE (effective border quarantine; test, trace isolate, and support). This will ensure that everyone is protected and make it much less likely that we will need further restrictions or lockdowns in the autumn.

The problems I have with this:

• How  are we going to get all adolescents vaccinated by this autumn?  We don't have any vaccines approved for adolescents, and there is far from widespread support for using vaccines under emergency use authorisation on younger children - and push back to that could be more damaging in the long term through fuelling anti vaccination sentiments, mercifully low in the UK compared to many developed nations.
   
• Assuming the problem with vaccine update is demand, we'd expect the current rates of sub 100k people/day to continue decreasing over the coming weeks and months.  How are we going to get vaccine uptake high against falling demand?  What are the limiting problems?  What more can be done?  
   
• No consideration is given to the constantly rising probability of a more vaccine evading variant appearing somewhere (likely by importation events) and the differences between natural and vaccine induced immunity.  This is an area riddled with uncertainty and difficult ethics I think, but they shouldn't pretend it doesn't exist - it makes the debate much more two sided than they present, even if there are intangibles.

At the start of this I strongly criticised some anti-lockdown posters for wanting the moon on a stick without a plan to get it.  I'd forgotten that turn off phrase until today.   

(The difference is the idiots last year had no idea the moon would crush them if they somehow got it on a stick, but the metaphors flips this time - having the moon on a stick as Indie SAGE want would be awesome but they still need a plan to get it ).   

This is where my patience runeth thin with Indie SAGE.  They have a lot of good points, but the demands they make are clearly politically very difficult and also look to be practically unachievable given the vaccination rate.

Issuing demands achieves nothing.  Instead - identify the key problems, focus on those, focus attention on those, solve the key problems move on.  A much more constructive approach, and many members of Indie SAGE have political and media access and power that they can bring to bear on the key problems.  Which right now is plummeting first dose vaccine rates.  

All good points. In the big picture the only major mistake in my view is around practicality of UK adolescent vaccination in that timescale (it is something that is happening in the US). Id missed that if reported previously. I've seen members of Indie SAGE on the news talking about worrying drops in vaccination rates.

There is simply no way the vaccination number drop is demand led as yet. There are plenty of young people who really want to get vaccinated. It's a supply and/or logistics problem, probably similar to those exposed on repeated Twitter posts. Even if it were demand led, at least those taking responsibility for their future health and helping the wider population would cut their risk of long covid as a bonus.

I'd disagree that making demands changes nothing. It gives concrete areas for opposition politicians to question government and to make alternative policy recommendations but it's less effective than it normally would be as our government doesn't give a shit about ignoring important questions. It's also not the only thing Indie SAGE do.

If hospitals start reporting cancellation of non urgent operations en masse these discussions get overtaken by events, as I doubt even Boris will be able to block extra restrictions. Our checks and balances on political decision making may have failed to be triggerred when they should according to the science on a few serious occasions but that position in the past always reversed when the inevitability of exponential data growth blew away any foolhardy optimism in dithering.  Boris really isn't Putin and our democracy has cracks and has failed at times but isn't completely broken.

"We don't have any vaccines approved for adolescents,"

Why not? My kids (16 & 17) have had both jabs and Pfizer-Biontec is approved down to 13.

New feature on the dashboard today is a log-plot option.

using this I noticed the rate of deaths where covid is on the death certificate hasn’t risen at all, whereas death within 28days of a positive test has. Ok, the cert data is laggier, but it seems obvious that the 28 days measure becomes invalid in the limit of high cases and low IFR. In other words people die of other things and if there’s a lot of covid about some of them will have tested positive in the previous 4 weeks coincidentally. True covid deaths might truly be very near zero..

Pfizer-BioNTech BNT162b2 has temporary authorisation from the MHRA for 12 and above in the UK.  I'd taken Indie SAGE's use of "adolescent" to cover down to ten or so, but perhaps I'm going too low down in age?  They could have been more precise in their language, perhaps I over-interpreted....

JVCI in the UK are not currently recommending immunisation of children under 16 except where there is extreme vulnerability, and 16-18 is only recommended if they are directly in a priority group or if a household member has compromised immune function.

We are I think due (or overdue) an announcement on those aged 12-17, but the news coverage suggests it will be a "no".

https://www.bbc.co.uk/news/health-57496074

In reply to Offwidth:

This is what I think needs urgent clarification - the Twitter threads, as interesting as they are, don't really get to robust evidenced status, and if the government are ploughing on with dropping the final restrictions whilst also failing to meet demand for first doses by a factor >> 1 that's a scandal IMO.  There's equally weak (in terms of evidentiary standard) anecdotal data suggesting there's not supply problems in various places but little demand.

If only there was a regular contributor to these threads working in a respectable part of the media....  

Interesting.  An analysis I keep not getting round to doing is a properly normalised comparison of the ratios for date X of:

• R_28 = (cases within 28 days of date X) / (deaths on date X within 28 days of a +ve test) 
• R_60 = (cases within 60 days of date X) / (deaths on date X within 60 days of a +ve test) 

Both deaths figures are available from the API although the later is  not well documented (the only googling of the API key a few months ago turned up a UKC page, where I'd hit on it with a lucky/informed guess...)

If the deaths data comes to be dominated by false positives, those two ratios should tend to the same value.  If Covid continues to contribute to the cause of death, R_28 should be larger than R_60.

Regardless of the details, remarkably low.  Various other countries have cases lower than us and turning to growth, with much higher deaths already; I see that as a readout of the efficacy (or otherwise) of their vaccination programs, and it doesn't look promising for their near future

Mainwood has concluded it's demand. Looking at the numbers in the usual places I'm struggling to conclude any different.

It does indeed appear to be demand. Which possibly isn't odd, really. You would only ever hear noise about people wanting but not getting. You wouldn't hear from the apathetic, and there aren't (comparatively) that many anti-. So I think we can explain away the anecdotal "I know of people who....". Maybe uptake is tailing off? Maybe the demand has been propped up so far by the tail-off in the stragglers from higher priority groups and we're running out of those? I dunno. Doesn't look like a supply issue anyway.

Edit: looks like it’s probably too soon to conclude anything from the very data due to the long lag. Maybe in a week..

WT that sounds like a very interesting analysis.. please..! ;D

Why do you think importation is more likely? Currently only Indonesia is reporting more new cases than the UK. I reckon the UK has more partially/fully vaccinated people than anywhere else at present, so logically our high infection rate and high vaccination rate makes evolution of a vaccine-resistant variant at least as likely as importing one.

Rest-of-world has a lot more cases than UK.

Well here the STIKO (that advise the government) only recommended for those in danger groups but that was soon brushed away, the doctors/health authorities said they couldn't refuse to give a fully approved vaccine for that age group (it is indeed 12 up). The current plan is to offer everyone in that age group by the end of Aug/before school starts again. The STIKO aren't amused at being overidden/pressurised by the politicians though but that's life.

At which point, the demand from Indie SAGE to wait for more people to be vaccinated makes little sense; as the counterpoint is a ticking clock on a variant with more evasion of immunity and/or more lethality in the young, at which point having a lot of young people around with no antibodies at all is a big liability.  I'd far rather they chose vaccination, but waiting for non-existent demand is not I think wise.

There is still a compelling argument to let second doses continue for another 4-6 weeks before dropping further restrictions, or even before allowing current rates to continue.  

When it's done I'm going to nail it to the door of an Oxford professor along with a post-it note explaining that this is the minimum level of rigour I expect for a blog post from someone who puts "evidence based medicine" in their job title.   

In reply to Toerag:

1. Simple numbers - the UK only has 7% of cases

2. The often suggested role of chronic infection in generating variants; with vaccine uptake very high in older and more vulnerable people in the UK, the fraction of our cases having chronic infections is very low (look at the almost absent deaths for example) compared to most countries where 20% or so less complete; they have plenty of vaccinated people to amplify a more vaccine-evading variant, and they have plenty of older people without vaccines to become chronically ill and generate variants.  I don't have a solid basis for this, but that's the way I'm leaning.

In reply to jimtitt:

Interesting, thanks.  I think a government override of the JCVI here would be an absolute shitstorm, but stranger things have happened.  There are strong arguments both moral and pragmatic that donating doses abroad instead of vaccinating children at home is preferable.  A busy time for the medical ethicists.  

But the UK is in a different place to e.g. Germany and the USA, as visible from 

I've been looking at 1st jab vaccine uptake from https://www.travellingtabby.com/scotland-coronavirus-tracker/ for a while, daily from 6th June. 

See graph at URL  https://ibb.co/xCrRPks

Uptake for ages 30-39 & 40-49 saturated/completed around 22nd & about 6th June respectively.

Uptake for 18-29 is rising slowly and I'd expect it to steepen as supply freed up around 22nd June when demand from age 30-39 tailed off. That didn't happen.

No sign yet of 18-29 uptake tailing off to saturation (of demand, i.e. completion) suggesting a limited capacity (seems unlikely) or limited supply of Pfizer & Moderna.

On the bright side, the longer the 18-29 uptake continues to rise without turning over to saturation/completion, the higher the uptake. 

Pity we don't have good data on Pfizer & Moderna deliveries.

CSV data below for anybody who distrusts links such as https://ibb.co/xCrRPks 

Date,18-29,30-39,40-49
20/05/2021,22,32,70
25/05/2021,22,36,77
26/05/2021,23,37,78
31/05/2021,23,38,80
02/06/2021,24,46,85
06/06/2021,25,54,86
07/06/2021,25,56,86
08/06/2021,27,58,86
09/06/2021,27,60,86
10/06/2021,28,61,86
11/06/2021,28,63,87
12/06/2021,29,65,87
13/06/2021,30,66,87
14/06/2021,30,68,87
16/06/2021,32,70,87
17/06/2021,33,71,87
18/06/2021,34,73,88
19/06/2021,36,74,88
20/06/2021,37,75,88
21/06/2021,38,76,88
22/06/2021,40,76,88
23/06/2021,41,76,88
24/06/2021,43,76,88
25/06/2021,44,76,88
26/06/2021,46,77,88
27/06/2021,48,77,88
28/06/2021,49,77,88
29/06/2021,51,78,88
30/06/2021,53,78,88
01/07/2021,54,78,89
02/07/2021,56,78,89
03/07/2021,57,78,89
04/07/2021,58,79,89
05/07/2021,59,79,89
06/07/2021,60,79,89
07/07/2021,61,79,89
08/07/2021,62,79,89

In the UK, the Health Secretary is legally obliged to take reasonable measures to follow the recommendations of the JCVI on any topic about which he has asked their advice. They therefore have a form of legal power (this is not covid specific). He is not allowed to pursue a policy which is counter to their advice, once he has asked for it. So vaccines for kids is not a decision that politicians here can take unless the JCVI recommend it. Fortunately, because I don't want Javid or Johnson brushing off their O-level biology.

What if he accidentally forgets to ask for the JCVI's advice?

Not that I can imagine any trustworthy Tory minister doing something underhand like that.

Did I say trustworthy, what was I thinking!

Demand limiting at around 50% uptake in the lowest age group certainly surprises me and would make the public health predictions wrong. ONS say "Over 9 in 10 (93%) of people aged 16 to 29 years have received or would be likely to accept a COVID-19 vaccine if offered"

How does Mainwood distinguish between those not wanting the vaccine and those not wanting to go to the vaccination centres where it is beng offered (more a logistics issue)?

By bet on vaccines is on either (1) logistics and delivery or (2) constrained supply, potentially to save doses for booster shots. I saw England data somewhere on twitter yesterday equivalent to what elsewhere posted above, that basically showed the same behaviour, ie doses not plateauing in young adults. I suppose alternatively it ma be that lots of people want it but not urgently and they are struggling to find a slot that is convenient. And as previously discussed getting your second dose brought forward can sometimes be a ballache. If they have spare doses why not make it 4 or 6 weeks now rates are flying up...?

If you'd been watching https://www.gov.scot/publications/coronavirus-covid-19-daily-data-for-scotl... you'd see it go up by about the same number as it usually does. No real change in supply.
Scotland and Wales report supply. They get a % share based on population. The number implies ~2.2m doses delivered cf <2m administered UK wide.

I'm as surprised as you to see it fall off, and I also struggle to believe it's demand, but what little evidence is available pretty much rules out a supply issue.

He doesn't. Why would 20-somethings find that harder than every other age group?

Supply into a country isn't the same thing as doing everything you can to make it easily available at the point of use though.

Because fewer vaccination centres are open now.

They opened lots on the first weekend 18 yos were eligible which then closed, as far as I can tell. The big centres round here are still running I think, but not every day and are alternating between vaccine types. Local councils are continuing to run walkin centres, which presumably makes up a bit for the reduced number of bookings, but they are only in quite constrained timeslots. It took me hours of searching to discover where Sefton are advertising theirs. If I was 18 and thinking I'd get one when asked but not inclined to go out of my way to find it, then I probably wouldn't have had it yet.

This is all just what I can decipher about the situation here from the internet, but I'm sure it's not unique. Other places in the country have much lower vaccine take-up on average.

I don't disagree, but if the explanation is that half of 18-29s can't get to a vaccination centre when 90-odd% of every other age group can, then we have some other problems to solve.

Just to be clear, I'm as bemused as anyone here, but none of the explanations fits.

a) supply issue
Well, scotland numbers show we've had plenty come in

b) logistics/delivery issue
We've suddenly forgotten how to give it out? Why now?

c) running out of people to jab
All indications/surveys/anecdotes suggest plenty of 18-29s want it.

d) GFBLs stole all the vaccine?
maybe....

Plenty of reasons:

- more likely to work in businesses that would expect you to be jabbed outside of working hours (bars, pubs ect)

- if they’re office worker, more likely to feel unable to ask for time off to be jabbed as more junior

- social life on a weekend that gets prioritised

- more reliant on public transport (compounding the problems above as it will take them more time)

Or perhaps they’re happy to have it, but won’t go out of there way to get it?

So we can reasonably reach a whole range of contradictory opinions on why the rate is so low.  Afraid the GFBLs aren’t cutting it though.

The question needs to go to relevant authorities promptly.  Hopefully it’ll catch media attention soon.  

https://www.theguardian.com/world/2021/jul/08/vaccine-uptake-coronavirus-en...
 

https://www.theguardian.com/world/2021/jul/08/vaccine-uptake-coronavirus-en...
 

Various other countries have cases lower than us and turning to growth, with much higher deaths already; I see that as a readout of the efficacy (or otherwise) of their vaccination programs, and it doesn't look promising for their near future

For Europe at least, some is left over from the last wave, ie France only just dropped below 1k ITU occupancy. 

Fancy that, thanks.

It’s a good point at the end of the article - and one your raised before but I missed acknowledging - that the number of young people with isolation orders is enough to delay a lot of first doses; perhaps 5% of under 25s?  Throw in inflexible employers and travel hassles and the knock on delays to a first dose could be quite significant perhaps.

True; hard to draw many conclusions at the turning point between waves.

Here’s one for you 

https://graphics.reuters.com/world-coronavirus-tracker-and-maps/

So far in Scotland the right number of appointments were sent out to do however many jabs were available. Hence you're not vaccinating the keen ones before the hesitant. It is the same random mix of keen/hesitant every day so you can get fairly consistent daily number of vaccinations with fairly consistent number of no shows and cancellations every day.

Hence largely linear rise determined by appointments sent out (limited by jabbing capacity or more likely vaccine supply) rather than demand until you run out of people to send appointments to. That's when saturation/completion reached.

Can anybody confirm if that's how it works (mostly) in Scotland for 18-29? The register to get on to the vaccination system didn't exist for ages 30 upwards.

On a positive note, I was idly thinking about the discussions on here recently about the likely dropoff in testing interest. Looking at dashboard data it's actually doing very well. Cases are just under half the January peak in England now (using 7-day averages), but the number of PCR tests has virtually caught up and still rising steeply. PCR positivity is much lower (some of the cases being from LFT). So it seems at the moment like testing uptake is very good. Maybe it's better in younger people because of better mobility and internet use than older groups... speculation.

It's not supply. Scotland publishes supply. It hasn't dropped. It isn't that. I don't know what words I need to use to get that across. It's on the government website for all to see.

Don't take this personally but in the last few minutes we've had speculation that they can't get to vaccination centres but can better get to testing centres....... um...

I really struggle to believe that the impediments to getting to the jab centre are enough more for the younger ages to explain the difference. Still don't know what the reason is though. I just can't see it being that. 30 yr olds also have inflexible employers etc. Are they twice as inflexible with under 30s?? I can believe there's a difference but I can't believe it's enough of a difference.

Edit: could be:

e) quietly building up enough of a buffer to start jabbing 16&17 yr olds before school restarts, in anticipation of a decision on doing so, and ensuring we can do it without jeopardising 2nd doses that would fall due in the gap before the August Pfizer order begins delivery. 

It's off the wall speculative fabricated horseshit from the depths of my imagination, but if it turns out to be that then..... I don't know how I'll feel about it. I think the GFBL thing is more plausible tbh.

I think we’re well beyond the point there is anything to gain from speculating.  My sense is towards demand problems not supply, but that doesn’t fit well with the engagement with testing.  Perhaps it’s distinct sub populations split by level of concern and flexibility of employment.

No point spinning round in circles when no determination can be made.  We could spin around in circles indefinitely with reasonable opposing arguments and go nowhere.

The question remains critical - if it’s supply, that casts a moral evaluation of current policy in a massively different light than if it’s demand.  

Can you point to published supply for Scotland? I remember early on they were slapped down on grounds of commercial confidentiality. If it's not a breakdown by AZ,  Pfizer and Moderna it's useless for judging supply for under 40s.

My children are aged 22 (double jabbed ages ago) and 19 (first jab on Monday).  I’ve asked them about their mates and plenty simply won’t take it. Quite a few of my contacts have young adult relatives - they report similar.  When I’ve looked locally, walk in centres are offering just about every time slot and location we could reasonably expect.  I think we are over analysing: a large  % don’t want it because they’ve been suckered in by the anti Vaxx hype plus some good old fashioned rebellion and wanting to be part of a social group.  No real evidence for this, just what hear on the ground and intuition.

I posted the link above. https://www.gov.scot/publications/coronavirus-covid-19-daily-data-for-scotl... 

"COVID-19 Vaccine supply data

As of Monday 5 July:

total number of doses allocated: 7,437,240

total number of doses delivered: 6,833,570"

The number just gets updated you so have to watch on Tuesdays at 2pm (or go trawling through the twitters)

What a surprise! Completely useless, it tells us nothing about how much is of the unused vaccine is suitable for under 40s.

As a point of interest from today’s data;

Germany: Doses given 79.7M,  Fully Vaccinated 33.9M, % fully vaccinated 40.8%

UK: Doses given 79.5M, Fully Vaccinated 34M,  % fully vaccinated  51.1%

So perhaps fewer supply issues in Germany.

It's the normal amount. So unless AZ have bizarrely increased production that we don't need or want by exactly the same amount that Pfizer have been unable to deliver, it's the same.

It's. The. Same. As. Usual.

Annoyingly the latest yellow card report is late, otherwise that would be another thing you'd have to find some excuse to dismiss.

https://statswales.gov.wales/Catalogue/Health-and-Social-Care/coronavirus-c...

Wales spares you the faff of looking up old numbers. Gasp. Supply not dropping.

I don't really trust the data.

The reason I don't trust it is that according to the dashboard they have actually injected just over 4050000 doses (4012000 as of 5 July), where according to the supply figures in your link which are not subject to validation, they have had fewer than 4 million delivered to the country as of 5 July.

The meaning of the data is different between vaccine types too.

Despite that I'm still inclined to believe what you and others have said that day to day supply to the country is not the problem but that's not the same as supply at point of use and availability of appointments.

Anyway, broken record. What Wintertree said a few posts ago.

(The likely cause of high testing numbers struck me by the way - there are lots of reports of high numbers of respiratory viruses being around at the moment and these will be prompting people to get tested, whereas in January much of the country had been in lockdown and other restrictions for a long time so there wasn't much happening with other viruses.)

What a surprise! Completely useless, it tells us nothing about how much is of the unused vaccine is suitable for under 40s.

You're better off looking at usage which has hard data of jabs in arms that day with breakdown by AZ, Pfizer and Moderna.

Fair enough. I don't really have a comeback to that. Stick with your assumption then, despite the only available information contradicting it.

Yeah, that raised my eyebrow a while ago so I did some digging:
https://gov.wales/covid-19-vaccination-programme-stock-and-distribution-20-...
ctrl-f for "may be possible to withdraw"

So I'm not overly concerned about correctness, my main metric is whether it's changing. Which it hasn't. So as long as they're the same wrong they're still implying no supply issue.

Interesting to note Wales has seen a lower uptake in the 18-29s than other age groups, and they've been offering to that group for a month now, so maybe despite all our hopes and indication to the contrary it could actually be a drop in demand. Sadly.

Yep.

James Ward now starting to hint at correlations between growth rates and weather. Thought you might like to know. Maybe get a "I thought of it first" flag printed and dated, quick?

The changes in rules for travel are going to make it more attractive for them to be double vaccinated.  It will be interesting to see how that plays out in the figures and may help resilve the supply / demand quandary.

Good point. The change in the rules could potentially be nudge tactics from the government?

Storey of my career, that.  It’s equal parts satisfying and frustrating to watch someone else get a run of good papers rediscovering something you did a decade ago but never got round to writing up…

There’s been at least one other stats punter wondering about the weather, dating back to the cold spell earlier this year, a woman on twitter.  I can’t recall their name - hopefully someone can put me right - or I’ll dig through the archives.

What I think they were missing at the time was that it’s not the temperature per-se that has the effect, but short term deviations from the short term mean it seems, which strongly hints at behavioural links IMO around meeting locations and/or ventilation.  I’ve been looking for the ideal Butterworth filter to reject the LF components from seasons and policy/vaccination and the HF components from noise.  

Edit: I was saving this plot for the next thread, but I'll put it out there now.

The MetOffice hadCET data is updated only at the start of each month, so I'm adding a provisional dataset from metcheck [1].

To my eye, there's an anti-correlation in the short term between the rate constant and the temperature, both deviating form their means in opposite direction.  As a result of a whole bunch of other reasons we ended up removing control measures and importing the more transmissive variant during the spring, so there's a correlation (not inverted) where both measures rise through the spring, but now it looks to be back to normal.  What convinces me of this is that the recent significant drop in the rate constant is synchronous across all English regions and Scotland, but "progress" through what we hope is an exit wave is not synchronous, far from it, so this is unlikely to be the rate constants tailing off as we run out of susceptible people.

Anyhoo, perhaps the blue provisional temperature curve gives a hint about where the rate constants are going. 

[1] https://www.metcheck.com/WEATHER/cet.asp

Have you considered using residuals from fitted curves to illustrate the negative relationship between temp and rate constant? It is rather confusing to have an obvious and massive (but almost entirely coincidental) positive relationship dominating the screen, while looking for micro-variation showing the opposite.

I do find this graph rather amazing. In ecology it would be too good to be true. But humans have a very different relationship with weather than other animals - the forecast and the habit of planning events. Which makes me wonder if instead of the actual weather, it is the messaging about weather that is important. BBC saying "it's going to be a great weekend" could have a bigger behavioural effect than the actual weather experienced.

Or maybe it’s not human behaviour at all, and it is the actual weather – could we be seeing short-term selection on the virus? A cold/wet tolerant gene and a warm/dry tolerant gene oscillating in frequency as the weather starts disfavouring the environmental survival of one and favouring the survival of the other. Is that remotely possible?

Could it be down to differential transmission due to the effect of temperature/humidity on aerosol lifetimes and dynamics? 

That's where I'm headed with the Butterworth filter; taking out the low frequency components is analogous to looking at the residuals from a low order fitted trendline , and taking out the HF components is analogous to the filtering done on the above plots to reduce day-to-day noise.

I got a bit further with it in yesterday's' downtime.  Putting both time series through a bandpass filter (5th order Butterworth) and then looking at the results of a linear regression between the signals.  The  heat-maps below explores the low and high frequency cutoffs for the filter.  One shows the correlation coefficient, on the other: red shows the correlation coefficient, green shows a null hypothesis violation for a linear fit and blue shows the standard error on the measured gradient.  

This is for an offset of 0 days between the datasets, a reasonable offset range should be explored to allow for lag in causality and detection.  On the TODO list.

My provisional take is that filaments of bright green/blue represent the dominant noise sources in the signals, and the islands of red between them are different correlations in the data.  Some of the noise looks to be fractal in nature which is pretty cool but probably just down to ringing in the filter design.  I need to talk with a more learned colleague about that, although it has no relevance to the parts of interest.

Too close to the diagonal and the passband is very narrow such that almost no real signal is going through, and the outputs just look like filter ringing, what I've labelled "ringing a bell on the plot.

There's then two main regions of correlation in the red block, one I think towards the low frequency end is the coincidental drift in baselines with the seasons and relaxation of restrictions, and the other is where the weather effects are.  Edit: No, the bottom left is just more bell ringing I think.

The line plot shows both signals filtered by somewhere eyeballed as around the peak in this region (LF and HF around 50 and 20 inverse-days respectively).  Most of the time the signals are anti-phase.  Context missing from this plot is a big change with the arrival of the Indian variant, and the easing of restrictions has been punctate rather than progressive which might ring through filters a bit.  The blue line is at y=0 and it's uncanny how often the two signals cross this axis at the same point.

In reply to minimike:

Therein lies the problem, there are so many possible mechanisms a credible investigation is a big research project in itself.  But, we don't see much seasonality in the exponential rate when restrictions are low (thinking of late last summer, for example), which hints against a physics level effect on viral particles IMO.

I thought that, but by the time they get the second jab the main holiday season will be over. Hopefully, some might get nudged by it. Perhaps the ‘don’t need to socially isolate if you’ve been in contact with a Covid case but you’re twice jabbed’ might nudge people, especially if employers start getting well pissed off with unjabbed regularly being off work. The cynic inside me thinks this is the gov’t thinking?

I suspect if you’re young and wanting to travel, you were first in line and will be pushing for the 2nd dose asap. There were opportunities enough for anyone over 18 to get a first shot long before  they could officially book one. 

Plenty of people in that age group go on holiday outside the school holidays.

I think it is a sensible " nudge". Nothing wrong with that. Besides you have to follow guidleines in other countries and places like Germany etc want you to have 2 jabs.

I'm feeling the voodoo now - looks uncanny as you say, but the methodology is just getting a little too opaque for me to properly grasp without working harder than I am inclined to. The Butterworth filters (whatever they may be), look like they might be a little too coarse (based on my gut instinct, which is of course, even more opaque!). On what basis do you select 5th order Butterworth? What do 4th order and 6th order Butterworths look like for example? 

Just Friday chat - I'm absolutely not asking you to produce 4th and 6th order graphs to justify yourself!

Any employer that isn't giving people the time off to get jabbed when they're incredibly likely to lose that employee to self-isolation or infection for a much longer period of time is an idiot.

Definitely. I know people scheduling their jabs to avoid side effects clashing with partying.

There's also a lot of hesitancy due to the general clotting / anti-vax / I'm not at risk of dying mentality.  It's quite telling that there's only capacity for drop-ins now we're onto the younger age groups.

It's the sharpness of the edges of the filter's passband in frequency space - a higher order means a sharper rolloff.  Perhaps it's - very crudely - analogous to asking you how you choose a basis function for fitting a trendline in the residual analysis you suggested.  

The results are very insensitive to the order of the filter.  I've figured out the bones of a null hypothesis test for the analysis to give the result some significance - it looks very promising; a bit too dull for the thread but I can PM you the write up if I ever get around to it.

I know it's not the question you were really asking, but here's a 4th order Butterworth bandpass filter with an active topology - an nth order filter is n filters chained together, so a 6th would look like 2 more.  Once upon a time I could have told you how it worked in detail, some part of my memory has obviously decided that's no longer important information...  (Well, the answer is just simultaneous equations, some slightly egregious assumptions about op-amp gain and conservation laws really...)

https://www.researchgate.net/figure/Butterworth-4-th-Order-Bandpass-for-ECG...

Already done; I tend to explore the sensitivity to somewhat arbitrary parameters in the analysis before I put any graphs out.  I don't want too have to post an embracing retraction because I'd been looking at an artefact of the analysis....  I don't tend to share it though as it's a bit dull and confusing in the context of the threads.

Thanks. You never fail to impress.

I think it’s the not at risk of dying point that’s driving it.

3 months ago the “vibe” in got from people in there 20s (I’m in my early 30s so I do still speak to young’uns) was that they’d get the jab to get things back to normal, or to prevent the old from dying.

Now that things are going back to normal anyway and the elderly have been double jabbed there’s less to motivate the young to get a jab.

The government needs to up their game in emphasising the importance of getting jabbed. Of course by saying that opening up is irreversible they’ve sort of shot themselves in the foot…

I've heard of people of all age groups through personal contacts displaying hesitancy. The youngest adults had higher hesitancy than most, due to what they saw as less gain for them and in young women particularly due to some fair fertility worries (due to poor government messaging). I've had very low reports of covid  conspiracy denial through the young. I also know of plenty of all ages who changed their mind and got vaccinated.. especially being mindful about holidays and the initial news from Bolton (of nearly all those who died being unvaccinated). Anecdote really isn't helpful right now in dealing with a known problem.

Irrespective, real analysis has been done (ONS and others) asking all ages if they will vaccinate. This means any 'demand' issues at current levels simply cannot be mostly due to hestinancy. Wintertree is right this needs publicity and investigation. There are all sorts of reasons especially inconvenient access to centres, and being positive recently that are simply not fairly labelled as demand.

Re: access to centres. Perhaps it might be because the north west has been badly hit, but local CCGs and partners have bust a gut with walk-in centres for over 18’s, multiple locations, varied opening hours. Screenshot of Oldham’s provision. Bolton is similar, Wigan’s not quite as thorough (3 sites).  
 

I’m not saying hesitancy/can’t be arsed is the only reason, but, given access appears good and no issues with vaccine supply, I reckon it could be the main one.

I've noticed that when the government wants to open up the BBC / Sky etc start reporting cancelled operations and happy people not wearing masks and when it f*cks up and needs to lockdown the BBC start showing grieving relatives and intensive care wards.   What we see is pure propaganda, the government either owns the media (BBC), is owned by the owners of the media (Murdoch, Barclay brothers etc) or is paying the media so much it gets what it wants (350 million quid in advertising spend for Covid).

I don't see why they can't vaccinate the same age ranges as the US.   The US CDC/FDA say it is OK and their opinion could be rubber stamped.  It would be a lot less risky than just letting them catch Covid which is the Tory plan.

I don't believe the slowing down is due to lack of take up .  I've got an 18 year old daughter and that age group are used to getting jags at school.  All her friends applied for their jags as soon as they were allowed and got them the first day they were offered.  I think the problem is they need mRNA vaccines for younger people and the great Tory vaccine procurement isn't as good as they make out.   

We have temporary licensing from our MHRA already as a result of our own robust procedures, there is no need to, and no benefit from, “rubber stamping” the FDA emergency use authorisation.

What we have, so far, is an active decision by the JCVI not to recommend vaccination of children with any of the products currently approved through temporary measures.

JCVI is not “Tory” and allegations of regulatory capture against it would seem both outlandish and unfounded.  Their decision can evolve over time, and I think we’re due an update soon.  

As I understand it:  Interestingly, the devolved Scottish government rubber stamp JCVI recommendations (declining to institute their own unit) but are not legally bound by them in the way England is, so as I understand it, the Scottish could decide to vaccinate down to the lowest licensed age (12 IIRC) with Pfizer/BioNTech but England can’t.  For now they seem content to abide by the JCVI recommendations.

It was your point that we don't have a vaccine approved for adolescents and it is holding up the vaccination program.  My point is f*ck the MHRA if the FDA say it is OK then go with that.  They are larger, smarter and have better access to Pfizer and Moderna people.  Most importantly they are run by Biden who is an OK guy but the UK regulators are run by the Tories who are scum.  It is naive to think the present Tory cabinet do not exert malign influence on senior public facing people in the public service who want to keep their job in the exact same way as Trump exerted malign influence in the US.

As a small/medium sized country England should just go with the larger countries that have the budget to do these things properly.

I didn't say the JCVI was Tory.  But I will now.  It is an agency of the UK government staffed by career people who want to keep their jobs.  The present set of Tories are evil in the same way Trump was evil.  People who want to keep their job will need to watch what they say.

That's interesting.  Personally I would prefer the Scottish Government to align with the FDA or EU Medicines agency as appropriate - i.e. mostly with the EU but occasionally if the situation in Scotland is clearly more similar to the US than the EU then with the FDA.   We are too small to need a full scale agency of our own and if you have a free choice why not follow one of the two leading agencies in the world rather than the English.    Right now my guess is when you get into the detail it isn't practical to do anything else but follow England.

Thanks for those. Oldham has a high population in BAME groups and more deprived areas than average for the UK so you would expect a bit more hesitancy than average. The data on NHS vaccination stats has 10,667 first doses for the youngest age group (not so low compared to other age groups, but of course 60% wider).. It's a pain they don't define the NHS data with a percentage. Oldham doesn't look much different from national averages on the ratios so that would indicate dealing with hesitancy is probably better than average.

You taking about the same FDA that approved aducanumab? Yeah, sure, they're impartial. No lobbyists pulling strings there.

Well which one?  They say different things.  And what happens when they aren't interested in a drug or vaccine that is relevant in the UK? Or when they don't consider UK medical facilities or public health situation?     I think you should swap polarity and just join the Tories - you'd get on just fine with their populism

Actually pretty much everything in your post shows you've done no research into how these bodies are comprised and have little or no appreciation of the scale or credibility of the UK biosciences sector.

You can foam at the mouth and chant "the tories" all you want but I'm not ok with you folding independent academics who are making good decisions in with them. Who do you get pissed off at when darling Nicola takes advice from scientists? Does she only ask SNP-voting academics? Have you checked?

Align with the EU.  Usually that will be just fine, it works for Ireland and Ireland isn't that different from Scotland and it used to work fine for the UK.

In this case the US FDA and the EU EMA are OK with vaccinating younger kids.  The English are the outlier so I'd just ignore them and follow the EMA.   An added benefit is Scotland following the EMA over the English agency would give the Brexiteers apoplexy.   

The US and a number of EU states have run out of adults who want the vaccine. They need to jab everything that moves to make up for that.

This shows something of a staggering disregard for the size of the biotech/pharma sector within the UK, and its global prominence.  Compared to the size of the sector, the cost of running the regulatory bodies pales in to insignificance, and there is no shortage of expertise.

Given recent developments stateside, I'm reminded how important it is that we maintain independent regulatory bodies.

• https://arstechnica.com/science/2021/06/a-disgraceful-decision-researchers-...
• https://arstechnica.com/science/2021/06/three-experts-resign-as-fda-advisor...
• https://arstechnica.com/science/2021/06/advocacy-org-calls-for-ouster-of-fd...

You are a Brexiteer in all respects but the country you obsess about.

The FDA is the agency for about 250 million Americans and influences very many other countries.

The EMA is the agency for 400 million Europeans.

So why would you not follow the advice from one of these global players? 

You've just had it explained above.  How about respond to

-That they aren't able to give advice in context

-That the FDA at least clearly has many failings

-THe UK is, despite your ranting, a global player in medical and biosciences.  The expertise here is a good as anyehre.

Because that is a shit metric for choosing who to trust. By that logic we'd follow the Indian regulator's decisions.

We heard from jimtitt upthread that the German panel did not recommend vaccinating younger children, but that they have been over-ruled by the German government - who have presumably taking a look at the vaccine uptake figures for adults in Germany, the R0 of the new variant and have consequently an einem Ziegelstein vorbeigekommen.

You still seem to be missing the point that in the UK, it is not the medicines licensing agency (MHRA/FDA/EMA) that are the reason we are not doing this.  It is an independent body that makes decisions on vaccination, the JCVI.  This seems analogues to the STIKO in Germany, who reached a similar decision to us, but unlike us they have apparently been over-ridden by their politicians, unlike here.

The situation seems to be a bit reversed from your take.

I think you need to learn the difference between the MHRA and the JCVI so you stop coming across a naive clown here. The MHRA relates to medicine and vaccine safety and so their findings are reasonably portable internationally, the JCVI evaluates the local and international information on a vaccine specifically in the context of the local situation and the demographics and medical history (in a general sense) of the local population.  It would make literally no sense for us to follow the advice of an agency deciding on behalf of a totally different population, medical system, level of adult vaccine engagement etc like the US.

I didn't  why we a government really. The Chinese one has 1.3b people  and is a global player. I think we should just let them govern us 

Tom, you've really derailed this hitherto useful and interesting thread with your political posturing.  Please return to your usual haunts where you and the usual opponents can shout at each other in a safe space?

This post is full of utter and complete shit. I don't like the Tories either but none of what you have written above is true. Fortunately, although some members of our current government are dishonest and verging on corrupt, we do have a lot of checks and balances in place.

I will do you the service of pointing out the current membership of JCVI from https://www.gov.uk/government/groups/joint-committee-on-vaccination-and-imm...

Even if you believe the rest of your own post, this should allay your concerns about the JCVI specifically. I hate the term 'career people' but if you meant people who are wedded to jobs for the government, well they're not, are they?

"Membership

Professor Andrew Pollard, Chair (University of Oxford)

Professor Lim Wei Shen, Chair COVID-19 immunisation (Nottingham University Hospitals)

Professor Anthony Harnden, Deputy Chair (University of Oxford)

Dr Kevin Brown (Public Health England)

Dr Rebecca Cordery (Public Health England)

Dr Maggie Wearmouth (East Sussex Healthcare NHS Trust)

Professor Matt Keeling (University of Warwick)

Alison Lawrence (lay member)

Professor Robert Read (Southampton General Hospital)

Professor Anthony Scott (London School of Hygiene & Tropical Medicine)

Professor Adam Finn (University of Bristol)

Dr Fiona van der Klis (National Institute for Public Health and the Environment, Netherlands)

Professor Maarten Postma (University of Groningen)

Professor Simon Kroll (Imperial College London)

Dr Martin Williams (University Hospitals Bristol)

Professor Jeremy Brown (University College London Hospitals)

Co-opted members (implementation matters)

Ms Anne McGowan

Dr Lorna Willocks

Dr Jillian Johnston

Dr Julie Yates

Conflict of interests

Any conflict of interests declared by members of JCVI are now published as an annex to the minutes of each JCVI meeting. The JCVI code of practice provides information on how conflicts of interest are managed.

In order to prevent any perceived conflict of interest it was agreed that the JCVI Chair (Professor Andrew Pollard), who is involved in the development of a SARS-CoV-2 vaccine at Oxford, would recuse himself from all JCVI COVID-19 meetings."

Applying nudge theory to the thread's current topics, updated England PCR rate constant plots below.  The weather one has an intermediate level of filtering between the raw data and the long term trendline on the other plot.

The last couple of days of rate constants have taken a big drop; more perhaps than fits in with my temperature related hunch.  

The question this frames is "are we there yet" in terms of herd immunity thresholds for the current restriction levels....

• The recent ONS result had ~90% of adults having antibodies by June 20th.  That leaves ~4.5m adults needing antibodies
• Looking at infections from one week before then to one week ago in the data (to give time for antibodies to develop) we've had ~0.2m cases, perhaps 0.5m infections with an optimistic hat on.
• In that same lagged period (to give time for antibodies to develop)., we've had ~2.8m first doses

Wildly optimistic estimate alert: So, if it's entirely uninfected people going for vaccination, we'd now have - with a very optimistic hat on - 4.5m - 0.5m - 2.8m = 1.2m adults without antibodies.  (I've wrongly assumed all cases were in adults here, but I'm already assuming a cases:infection ratio and that's the more egregious assumption).  That would leave 97% of adults with antibodies which should stop things dead.

But, in reality...

• Some fraction of the people in the ONS survey with antibodies (up to ~25%?) had them from infection and then went for vaccination, meaning that they're being double counted in my estimate above.
• Some amount of infection is in the vaccinated, and a smaller amount in the previously infected.  These are being double counted in my estimate.
• Some fraction of the people receiving vaccines won't develop antibodies and therefore presumably protection; a small fraction as that's mostly old people who already have high uptake
• Maximal protection against infection and transmission requires both doses, with a lot of those still lying in the future (around 10m).  That probably makes a correction of about 2m to my estimate above

I'm not sure thinking that through has really improved my understanding of where we are!  Just adding noise to the discussion....  

If the exponential rate constant remains low for the next week or so, I'll not be as concerned about the likely dropping of all restrictions that's coming up, but I'd still rather that it was a progressive dropping to give time to keep evaluating hospital occupancy levels and bide time accordingly.

The absolute number of cases/day is still rising in England; it looks suddenly to be falling in Scotland, but, as I've said before, I think the uneven nature of vaccination means we're subject to larger scale outbreaks and perhaps we're seeing the collapsing tail of a football related outbreak that's masking a more gradual rise.  

The slowdown in exponential rate is correlated across all English regions and Scotland so - given the de-synchronised progress between areas - I think the weather still has a role in this drop.  Looks like a great week or two ahead; if the weather really is depressing rate constants, dropping all restrictions and then getting a spell of crap weather could cause a bit of a shock - although I'm not sure the weather effects are going to translate to transmission in nightclubs and so on...

Right now I wouldn't be surprised if we see growth really start to level off, although I'm 50/50 on that or continued growth.   Both seem reasonable against the conflated snapshots we get in to the data.  Only time will tell.

Would it? With the new, ever increasing HIT? What % of people are susceptible and children? I think that's the same question as "how many fingers am I holding up?" and at the same time can be answered by "you'll find out real soon". But still.... Big shrug from me.

Edit: and a lot of them antibodies are the first dose, 30% effective ones remember. I agree with your working but I would start from different assumptions. Pulled out of my ass, but numerically different.

Depends what I meant by “things”.  Hospital admissions or ITU admissions in the under 60s would hopefully be more or less eliminated at that point, circulation of the virus - I’m not sure anyone really knows at this point, but I doubt we’d see fast doubling times any longer.

Fair point.

Kids though.... There's a lot of them. They put herd immunity out of reach don't they? Without the live unattenuated vaccine at least.

In other news, have you seen the new logscale button on the dashboard? Used it? Does fun things with negative numbers. Good code that.

Also does kind of show sub-exponential growth, same as all the other plots of the same including yours. Or at least a move to a new, shallower gradient. Hard to say which with ~4 days to go off but I'll take either.

https://imgcdn.ukc2.com/i/370077?fm=jpg&time=1625825426&dpr=1&v...
 

As you know I’m a huge fan of all this, but In the spirit of open review I’m going to say I’m nervous about this analysis (realise you’ve caveated it heavily 😉)

I don’t think it’s an issue with filter choice or order, but rather with the frequency cutoffs. The filtered signal looks worryingly monochromatic (single frequency) to me. Check this isn’t the ringing frequency of the band pass... easy check is to shift the band a little and see if the signal shifts.

also, looking at the filtered on the raw, I just don’t see all the same features so it’s failing the smell test for me (although I can’t smell much since last week 😂)

Mike

The Laphroaig I hope and nothing more sinister.

Good comments on the analysis.  

If you look at the heat map, the island of correlation I circled called "Weather effects" maintains a high R value to a much higher cutoff frequency.  

Tine for a sanity check as you rightly suggest.  Three plots below

1. A wider pass band (4x high frequency cutoff)
2. An alternative implementation plotting residuals (bottom) determined by subtracting a mild polynomial filter (removing day-to-day nose) from a harsh polynomial filter (long term trends only),
3. The original narrow pass band.

My take flipping through them is that the two signals are generally anti-phase on all of them, and the red curve is above the black on (2), it is above the black on (3), with better removal of baseline drift but less good preservation of amplitude.

Filtering erratic time-series data is always a slippery job, especially when you start staring at the plots, I get an effect like semantic satiation after a while where the data looses all meaning.  

I put together a null hypothesis test for this this evening; do the same correlation coefficient measurement for all pass bands for every year for 2013 to 2021 inclusive, and calculated the mean and standard error on the mean for each value in the matrix across all previous years, and then calculate a significance value for each element in the matrix based on how many SEMs is from the mean of other years.  The area I've been drawing pass bands from and circled in the earlier plots has way higher significance.   I'm not a great fan of the magnitude of these values - they seem very high so I've got some more thinking to do.  So, I think the effect can be justified as perfectly real, and it persists even when the data is filtered beyond recognition because it really does live in quite a narrow passband between the seasonal/policy effects and the noise.  

Ok. Some evidence but the absolute phase of the oscillations (peak positions) is changing with the HF cutoff, which worries me more. Now change the LF to 8.5/yr and show me the anti correlation is still there...??

im lost on your null hypothesis I’m afraid.

sadly not.  I can’t even taste the laphroig I just put in my hot toddy. ;(

seems the military grade viral load my 6yo delivered (amplified by my 2yo) was too much even for double dose vacc-schprung durch technic.

Well, that sucks.  Best wishes for all in the minimike household.  Waste of Laphraiog though if the virus has your olfactory nerves...   I was going to have some with this evenings movie but my past self is a thoughtless b*ard.

I've made some pretty drastic changes this last year to get away from peer review, yet here I am.  Seriously though, I appreciate the perspective and challenges to the analysis, thanks.  Plot below as per your request.

Plot below with the LF @ 8.5/year.

If I can explain the null hypothesis I might make a convert...

• Top-Left:  Correlation coefficient of filtered temperatures vs rate. constants for a low (y, /year) and high (x, /year) frequency bandpass.
• Top-Right: The average correlation coefficient from a set of matrices like the one in the top left where we expect no connection to the rate constant.  These matrices are formed by using the temperature data from 2013 to 2019 (inclusive) which should have the same general seasonality but no correlation in the fine details.  We see there's stuff happening along the diagonal which is I think ringing of very narrow-band filters when excited by the data going in, which I think is part of what you're concerned about here?  But there's no big negative correlation in the area I pegged as "Weather Effects?" in the low frequency corner.
• Bottom-Left: The difference between the matrix for 2021 and the mean of the null hypothesis matrices across the null hypothesis years
  • This shows that there's a lot more correlation between data for the band-pass filters in the lower left corner of the 2021 plot than the mean for the null hypotheses correlating with temperature data from other years
  • The values are all more blue, showing that it's an anti-correlation (higher temperature deviation, lower rate constant deviation)
  • The red island is an area of correlation in a much higher frequency bandpass - but I think this is basically noise - see below.
• Bottom-Right: The absolute value of the bottom left matrix divided by the standard error on the mean across the null hypothesis matrices.  
  • This shows that the higher correlation in the lower left of the 2021 data is higher than that of the null hypotheses in a way that's got statistical significance.
  • There's nothing like as much significance to the red area - I think this is higher frequency noise stuff?

Now, if you look at the bottom right plot, x=~20/year (high frequency cutoff), there's a bright white band showing high significance to the correlation, and that extends vertically across low frequency cutoffs from ~7/year to ~14/year.  That shows that the statistical significance of the anti-correlation is highly robust against the choice of low frequency cut-off.  The area of high significance is quite with in the 'x' axis as well, showing it to be robust against the high frequency cut-off.

The peak positions moves around with the band characteristics, but that's to be expected I think; the anti-correlation is strongly preserved for a wide range of bands.  I want to do a similar analysis for the low and high values using SG filters as per the middle plot on my previous post.

Edit: I think I prefer the trendline residuals plot Wicaomi suggested however, as shown on the middle sub-plot of my previous post.  It's much easier to explain to a wider audience, and it seems to strike a better balance of preserving key features of the input signals vs stripping the data down to the bare minimum with a side order of Fourier weirdness, disarming skeptical readers...

Re vaccination rates - we’re at 87% first jabs and I’m not sure many people expected such a high %. Not surprising that it’s running out of steam and in a way it’s great that numbers are still growing. I suspect we might not get much above 90%.

I wonder how many of them are from England?  Actually never mind England, I wonder how many of them are from the south of England.

Oh look.  11 from the south of England.  2 from the midlands of England, 2 from the Netherlands and 5 no stated affiliation so no way of knowing where they come from

London rule.  People recommending their mates is how you get a committee with that kind of geographical spread.  I wonder how many of them have at some point in their career worked with the Oxford group that did the AZ vaccine.

I never said they were corrupt but unless you are English I see absolutely no reason to give this group of southern english people any more credibility than the EU or the US regulators.     If I said Scotland should follow South Korea (or any other advanced medium sized country apart from England) when it could follow the EMA or FDA you'd think I was nuts.   

Your whole argument is based on English (and London) exceptionalism.  The reasonable thing for a small country to do is to largely follow the decisions of the largest and best funded regulators and that would be the EMA and FDA.

The link with temperature is interesting but as climbers we know that it’s as much if not more about sunshine and wind speed and direction as about temperature. Ok, clearly people looking to foot the pub don’t look at the forecast as closely as someone looking to do some conditions dependent route but they will still take into account the overall weather - including, and perhaps most importantly, rainfall.

The other thing to bear in mind is that in the summer months the temperature differences aren’t as important. I get your point about deviation from the mean but I don’t think there’s going to be a massive difference in the number of people going to the pub at 18C vs 22C.

You are missing the point, upthread someone pointed out that Scotland are not legally bound by JCVI.  The English JCVI is too slow given the Tory time table to let this thing go ballistic and shouldn't be given any special status by Scotland, it is just one of ten or twenty similar agencies in medium sized countries.

The situation is that the EU and US regulators (and actually the UK medicines agency too) think the vaccine is safe for kids.   The Tories are setting out to create a huge wave of infection which Scotland probably won't be able to protect itself from.  The kids either get vaccinated or they catch Covid without being vaccinated.  Their parents should have the choice to get them vaccinated first.   

I see that list and see lots of professors. That generally counts for something.

As others have said, you make many good points but their effectiveness is lost due to some of the tosh you come out with.

I think a good argument against vaccinating children is that there is relatively little benefit for them, whereas the doses could be used for adults in other countries. The counter argument is that our current policy increases the risk of a vaccine resistant variant, so we need to keep cases in check and one of the ways to do that is to vaccinate children. So I can see it both ways. I make no comment on the scientific case as I’m not qualified to do that. These are just very general observations.

I was about to start a long reply but it’s 2am and I can’t be bothered really, this virus has knocked me for 6.. 

anyway, hope I didn’t overcook it with the peer review! Might have slipped into default mode there and forgotten this was UKC! Oops..

in short, the plot I asked for doesn’t actually allay my concern, probably because it was a poor request.

the null hypothesis makes sense now (thx) and I like the analysis. However..

I think the problem here is you’re driving a set of similar IIR band pass filters with an impulse, which is anti-correlated. Once you’ve done that they will ring at the impulse frequency provided the resonance is close enough. This explains the horizontal ‘stripes’   in your correlation plots I think. (LF cutoff more strongly impacts filter impedance matching to the driving impulse).

This says 2 things to me..

1) there IS some genuine anti correlation in the signal at around the passband frequency.

2) processing this with IIR spectral domain approaches is fraught with difficulty. 

this leads me directly back to plot 2 of your previous post with the SG filters... oh, wait! ;-p fools seldom differ, eh?

^^this.

I totally buy that approach. In fact it shows the series of anti correlated fluctuations which is driving the filters above. However without the ‘Fourier weirdness’ it does not continue through the whole data series in the same way. The peaks line up much better with the data (as you’d expect)

in some senses what you’ve achieved here is a kind of template matching. You’ve identified the driving impulse very strongly but at the cost of knowing it’s temporal duration and dynamics which are driven by the filter more than the signal. It’s a transform limit, if you will? No actually, it literally is.

interesingly there was some scepticism back in 2015 over the initial LIGO/VIRGO results because they used a similar approach (chirped IIR filters essentially) to template Kerr ring down in the black hole mergers. So they were in a situation where a noise correlation in the wrong place could set their template filter ‘ringing’ and synthesise a fake GW event. Of course if turned out they were being cleverer than this and all was good, but it was a valid initial concern (damned peer review again!)

ok, if that’s my shortened covid post I need to get out more. Oh wait.. I can’t.

Late edit: some insomniac reading about SG filters has me learning that they can be considered to be FIR filters like anything else. The ‘frequency response’ is determined by the interaction of the convolutional window (no. Of points) and the polynomial order. What order are you using? I think this is important as it potentially leads to the same Fourier issues via a (rather subtle) back door. Having said that I am happier that I can see the features the SG analysis highlights, so I’m less concerned. 

if you want to invest (waste?) any more time on this I’d suggest a sensitivity analysis on the SG using both window and order.. I think uou already proposed doing it for window? Anyway, feel free to ignore me, we’re WAY past covid plotting now..

....with new football outbreaks from Wednesday's game and this sunday's game to come....and they're going to be pretty big.

Down here the big factor is rain - people are lots less inclined to go out when it's tipping down. Wind, not so much until it hits F6.

My stats show it levelling off. Footy will put it back up again, stories of NHS running out of staff due to S-I will put it down again.

yeah, completely impossible. It shall forever be a mystery. There's absolutely no way of knowing. There definitely isn't a website where you can just type someone's name in and get that answered. 

https://www.google.com/search?q=Dr+Lorna+Willocks 1

Some next-level ignorance going on in your contributions here. Impressive, in a way.

yeah, completely impossible. It shall forever be a mystery. There's absolutely no way of knowing. There definitely isn't a website where you can just type someone's name in and get that answered. 

https://www.google.com/search?q=Dr+Lorna+Willocks 1

Some next-level ignorance going on in your contributions here. Impressive, in a way.

Is anyone taking minutes? Should be writing this down for next time chlorine washed chicken comes up.

Of course it does.  Nobody is saying they are not respected academics or that their opinion is not far more valuable than that of politicians.

The question for Scotland, should it be allowed to take its own decisions, is why go with the English set of professors instead of respected sets of professors from the EU or US.   When you add in that the Tories are about to let go all control and it will almost certainly flood over the border to Scotland and the English JCVI is still deliberating the obvious thing is to go with the US/EU advice and offer vaccines to kids at the option of their parents.  

That is assuming there is enough mRNA vaccine available to start vaccinating kids.  We had this situation before with face masks.  They delayed advising people to wear masks when there was a shortage of masks so as not to create competition for supply with the NHS.  Maybe they are delaying approving the vaccine for younger kids because there is a shortage of mRNA vaccine and they don't want to create competition for supply with young adults and older people.   

My thoughts are within a couple of years the Brexiteers will sign a deal that means we will be eating it in the UK but they won't be eating it in the EU.

Scotland should break away from England and follow the EU on every regulation where it has the power to do so.   It will hasten the break up the UK and make it simpler to return to the EU after independence.

A self-hating Scot recruited by Tories to kill children. It's so obvious.

The thing is, there are good arguments for an international approach in a lot of this. Just not always, and not for TiEs small minder nationalist obsessive reasons. 

Yeah, if only we had some advisory panel of experts who could help us decide which countries we should align with....

Please could you desist from bloating up this thread with your Scotland stuff? Start another? Thank you.

The situation with the STIKO is a bit different to the UK because while their decisions aren't directly legally binding they have a bearing on the decisions made by the government, primarily on who is paying and who is responsible for a poor outcome. But application of policy is up to the individual Länder.

Once the agreement on common vaccination policy lapsed (based on age bands) many made it open and vaccinated anybody who wanted it and the STIKO recommendations are somewhat open. They recommended vaccinating risk groups but also allowed any children to be vaccinated if it was their (or their parents) specific wish.

Bayern is starting vaccinations in the schools BUT only for over 18's, the problem for younger kids is the informed consent from the parents which is why there is considerable political pressure on STIKO to change their recommendations. One problem being that STIKO is somewhat slow to respond, their recommendation being made before Delta became a problem and the effects it has on long-Covid where known, the politicians and health authorities are expected to react within days or weeks, not months.

So lander to lander quite a lot of variation in implementation? I was pretty surprised when you said your daughters were vaccinated since a lady in her 50’s in Hamburg I know had not been offered her first dose a couple of weeks ago.

Tom, re JCVI Scotland already can make its own choices - Its England that’s legally bound to follow them.

and in terms of Scotland being flooded by a wave from England, the most recent ONS data shows Scotland with a higher per capita rate than England .

Scotland peaked a week or so ago at 3454/day (7 day rolling avg) and has been falling, now at 2999/day.  

England is now at 30k infections/day (so scaling for size about the same as Scotland right now) but it is rising where Scotland is falling.    Various Tory ministers are talking about letting it go to 50 or 100 or 150k after their freedom day.    150k/day scaled for size would be 15k infections/day in Scotland instead of 3k (and since it is falling fairly fast at the moment maybe 2k by the 19th). 

'Wave' is a completely reasonable description of what the Tories are planning on releasing.

Even more crazy is to spend 38 billion on a test and trace system to monitor Covid and then before you do an experiment on this scale tweak the app and charge for tests to make sure that you don't get data that you don't want to see.   It's like the guys in CERN spending a billion on a  fancy muon detector and switching it off before firing up the collider just in case it finds something that conflicts with their theory.

Do bear in mind the 38 billion is mostly testing - not the app.

I think tweaking the sensitivity of the app is not unreasonable - as a greater proportion of the population become vaccinated AND virus characteristics alter then it’s bounds will need revising. 
 

Since the App has not ever really been used to drive policy, I don’t see this as a problem. I assume if the original app plan in which data was centrally gathered for epidemiological purposes had been adopted (I recall you vehemently opposed it?) then it would have been possible to maintain any useful stats input and tweak advice to contacts.

(Dr.S currently waiting until midnight on Thursday to be released from advisory isolation)

You really do only see what you want to see, don't you?

The app is pinging more people because contacts are one generation ahead of infections and infections are rising.   Turning down the sensitivity is like turning off a smoke detector because it is beeping all the time.

I absolutely do not see what I want to see.  I want to see Scotland back in the EU, dealing with Covid in roughly the same way as Germany or Ireland, and with the border to England firmly closed until this Tory madness is over.

Yes it’s ‘pinging more ‘ because there is more disease about - but as the disease and population changes the desired sensitivity of the app should change as well - in either direction depending on the evolving disease picture.

The app changes are frustrating because of poor communication, and the obfuscation about the advisory vs non-advisory nature of the alerts - not because of the intention to alter the sensitivity.

At least the planned changes to management of traced contacts is signposted well in advance, for the 16th of August, but that of course adds its own frustrations for folk traced before then.

one of the great benefits of devolution is that different bits of the U.K. can try different approaches - are the Scottish government contemplating any changes to the implementation of their app?

Now you know that the JCVI has no statutory basis for providing advice to Ministers in Scotland, I presume you've changed your mind, and the vitriol you poured forth on the JCVI and their 'slow' decision making, can in fact be directed closer to home.

If this is the level of insight and review you can come up with when self isolating with two small children and Covid, I should probably just hand the job over when you're better... 

Makes a good distraction from the "freedom" ruck on the thread...  Also, it's much appreciate - moving away from the world of peer review into new areas I get much less exposure to review, so it's good to get some critical analysis on the way I'm approaching something, even if it's on the hobby side of my activities.  Thanks.

Yup.  I think it's built a very sensitive detector for anti-correlation but it raises a set of difficult concerns, especially over noise. 

I hope you had the requisite "WTF" moment at finding out the SGs can be a Fourier domain convolution...  I should have said in the first post, they're first order filters.  

What I like about the SG filters is that the polynomial fitting gives a time-domain basis for filtering at the edges of the dataset where the convolution fails, removing a set of arbitrary and often inappropriate choices about handling that in the Fourier domain.

The combination of a high- and low- filtering with SG becomes a Fourier domain bandpass filter, but with softer edges than a Butterworth; removing some of the sensitivity which is both a pro and a con (considering noise).  The big benefit is that you can explain the whole method to people in fields where SG filtering is known but Fourier theory is almost entirely absent.

That's the basic plan, although order and window aren't so independent when you think about the number of free parameters in the fit and the number of data points being fit to (in a time-domain mindset).  I also need to explore this over a reasonable range of lags from the weather to the rate constant, as it's based on detected cases and not infections.

I think the LIGO link is really insightful, wish I'd had it.   The null hypothesis I've picked reassures me that noise triggering an over-sensitive detector isn't likely, but I have a much worse handle on my noise statistics and sources than LIGO.  (I sometimes think - and this is praise not criticism - of LIGO as a project to study noise that happens to involve some interferometry.  If we are living in a simulation they'll be the people to spot it...).  The noise in the cases data just isn't very sane, it has temporal correlation which itself is not sane but depends on a modulo-7 effect.   

It's never fully wasted if I learn something...  I think going down the SG route a bit is the better way forwards as easier to present onwards, and it's easier to address concerns about noise ringing the detector.

Logically, if the disease gets more transmissable as it has with Delta then the app needs to be more sensitive rather than less sensitive in order to provide warning of potential infection.

It is pretty clear what the Tories want is that people should go to work even if they are potentially infected or actually infected but without serious symptoms.   They want the app less sensitive because it is getting in the way of their policy by acting as it should.

I have no inside knowledge of what they are going to do with their app.

From Nicola Sturgeon's latest statement she clearly isn't going to follow Johnson and remove all restrictions on 19th July.   Masks and social distancing are going to stay in Scotland for some time.

I didn't mention it in the hopes it won't happen.  The scale of school disruption going on now is the last thing anyone involved needs, and I'm just going to end up ranting.

Good point; as Wicamoi noted up-thread however correlating rain is much harder than temperature.

Yup, so much uncertainty about what's coming with both great trouble and perhaps a great improvement in our situation both within the bounds of that uncertainty.  A big shift in the media today over the wisdom - or otherwise - of dropping all restrictions on the 19th.

I did!! Thankyou for the opportunity.

I don’t understand what a first order SG is though. I’m talking about the order of the polynomial, so first order (linear) would in fact be linear LOESS, no?
 

edit: Ah, ok there’s no distinction.. they’re the same! Ha. 

Im used to thinking of SG as at least quadratic local regression, but it’s still SG if linear too. Semantics

‘Dirac and Fourier went to sea on a beautiful sinusoid..’

https://www.ldeo.columbia.edu/~richards/webpage_rev_Jan06/Ch3_FourTrans%26D...

Bravo you two - if only all peer review were this civilised!

One of those moments I contemplate how basic my maths is, and wonder how that limits my insight in to things.  All this was figured out before computers as well.

Yup; intending as I do to look at higher orders it makes sense to stick with the name even if it's just local linear fitting for now.

For interest, filter responses below for plots (2) and (3) in this post - https://www.ukclimbing.com/forums/off_belay/friday_night_covid_plotting_33_... - I measured these with white noise and the same code as used for the data.   The Butterworth cutoffs came from finding the best significance against the null hypothesis in the 2d parameter space, the SG ones from me eyeballing the data and picking what seemed sensible to the different timescale.  I'd say given the good match of the SG filter to the data when you look at it, this is a sign that the notch filter is doing the right thing, just a bit more mysteriously so.

I wonder if any of the LIGO people would be interested in putting some smarts in to this...  How to evidence the significance of the temperature correlation in a way that is scientifically robust is not as simple as I first thought... Or I've just over-complicated, drop the Butterwortth, optimise the SG approach, keep the Fourier interpretation out of the write up and hope the null hypothesis holds epidemiological water...

On my reading list is a biography of Oliver Heaviside.  He was a fascinating person who contributed massively to modern Physics through force of character and inspite of significant challenges.

He also invented the Heaviside step function, few people remember him for this compared to Dirac, which is a shame because the later's delta function is purely derivative work.

.

.

.

.

.

Boom-Tish.

You mean a conversation rather than a fight?

Here’s your coat.

I suspect it starts with the fact we actually read what each other writes.. 😂

Nice.. what do the temporal impulse responses look like?

edit: are there real live LIGO people on here?! 

but that’s more transmissible compared to original - the populations increasing resistance also plays a part as does the changing out come of infection.

well that may be the case - and it could be sensible to do this as the disease:population interaction changes.

My case in point, one of my work colleagues had symptoms of a cold. He ran a PCR and it was positive. Through a combination of our internal contact tracing and NHS app we sent home c 15% of our work force.  Most of that 15% are double jabbed, and as yet no one has developed symptoms or shown positive LF results - is our action still proportionate? It would have been in February this year, is it now?

But even in England not all restrictions will go at that point - you will still be legally bound to isolate if you test positive, and will either have to isolate or get a test if you are a contact.

I’ll have a look later.  We’re off to hunt more alpine strawberries now…

I meant in offline life in that comment, but now you mention it, there might be one on here IIRC but I’m not sure…. 

Ooh yum!

Plant some in your garden, they spread like mad.

The SG looks like what you'd expect from a quick think on it, the Butterworth rings with a phase shift dependant on the filter parameters.  

I definitely think the Butterworth gives more sensitivity - both good (to data) and bad (to noise) - which, given the lack of a proper noise model for the rate constant plot suggests the SG approach is a safer bet, and why the features in its filtered data align consistently with the input data set.

In reply to Toerag:

I've had limited success, no doubt linked to the plague of rabbits...

Can't  get rid of the damn things, and  berries are tasteless.  Good blueberry harvest today however. 

Thanks. For me that plot settles it. SG has strongly constrained ringing characteristics that are time symmetric (in fact its frequency response is the result of this as it’s a TD specified filter). The BW is typical of high performing FD filters in that it imposes a TD structure on any impulse. Obviously the data aren’t driving these filters with impulses but they aren’t infinite sinusoids either.. wavelets again??!

no, I think it’s already enough. SG is good because we are more interested in tightly constraining the temporal response than having a narrow and sharp frequency band.

Idle pondering, perhaps:

At some point the number of people under isolation orders is going to limit the exponential rate, not just because of exposed people being quarantined but because so many people without the disease are stuck at home they're not out there catching Covid. I wonder if the dynamics of this settle to a steady state model or to a two-state model pogoing between periods of high spread and high isolation?

In reply to minimike:

Right, I'll fire up the time machine and use Wicamoi's suggestion instead of my original take, delete all the confusion and make it look like I backed the better methodological horse from the start, thanks.

One of these years I'm going to sit down and learn about wavelets...

I'll bet you the Butterworth produces a more sensitive detector though; but perhaps only if you already know what you're looking for which invalidates it a bit here.  Although if you were then examining other nation's data against their weather patterns (it suddenly occurs to me that that's a powerful way forwards...)...

neigh problem ;D

consider that a man with a hammer is a very sensitive nail detector. But maybe not so specific..

edit: on the plus side, I’m feeling much better and can smell garlic and marmite (weakly) so looks like I’m on the mend ;D 

thanks for the mental chewing gum, I’m sure it’s helped the convalescence! 

Was considering chiming in here but didn't want to piss on your chips when you were both having so much fun learning; how about just plotting the data, and filtering with mk1 eyeball, for multiple countries? I think that was just mentioned, but I'm not gonna lie, I've only skimmed the last day's posts.

What was under your nose at the time?

Anecdotal, but during the last couple of weeks I've met quite a lot of people who are isolating. Daytime supermarket queues seem to be a good place to meet them.

It’s a decent idea but there are issues, like the unknown reporting lags from each country and the reliability of case numbers. Also the differences in cultural and underlying social dynamics which are likely at play here..

I wonder whether I can detect a whiff of patronising sarcasm in your post. Hopefully that’s just the lingering effect of covid on my olfactory system..

That's how this started - just with the English data; what I've been looking for is a way to turn it in to a testable hypothesis so that it can demand serious epidemiological attention and not just be a random person pointing at curves and leaning on wishful thinking.  I think that using a bandpass filter made from two SG filters and using the "past years null hypothesis" I can get that, along with plots showing the workings - to bring the reader along - and the effects that are visible will be compelling for people who don't think in Fourier space.

In reply to minimike:

Prevalence of domestic and workplace HVAC is probably another big factor.  The data might either be conflated by the social/cultural stuff, or if you can identify the differences between nations it might actually reveal which factors are the important ones.  That's well past a hobby level investigation though...

Good, good.  On the ++ side, you've not got a triple-whammy of antibodies any perhaps aren't as at-risk from the next variant to come along as you were a few weeks ago...  There must be a market for blood samples from the infected immunised to see what it looks like...?

He wants my blood. I knew it..

Sleep all day.  Party all night.  Never grow old.  Never die.

Explains a lot

No, didn't mean it like that. Sorry if it came across that way.

I like a good sav-gol filter. It's my go-to for taking out higher frequency stuff that doesn't play well with finite elements.

just wondered if the quest for the best analysis might be the perfect vs the good in this example, and whether it should be necessary given how visible the match is in the raw. Should have done you both the courtesy of reading properly before passing comment though. 

Fair enough, thanks for the apology, although seems not required, ;D

 Yes it’s admittedly been an academic exercise for the sake of it. I’m not sure any of these analyses strengthens the case for the anti correlation but it’s always good to know what biases your analysis methodology can introduce! 

The app should be calculating the probability of an interaction leading to infection and comparing that to a threshold.   There should be several variables feeding in to the probability calculation and now that Delta is the prevalent strain one of them should be increasing.   The app should know the phone owner's vaccination status and whether they already had Covid so it could reasonably have another variable which reduces the probability of infection based on that data.   You could also argue that the threshold should increase if the outcomes are less severe and, again, you could modify the threshold based on the phone owner's age and vaccination status.

I don't think any of this is happening.  It is far simpler, the app is doing its job and pinging large numbers of people because there is a surge of Covid infections and the Tories don't like that so they are going to f*ck with it in a brute force way until there are less pings.

Well, despite large numbers of people getting pinged and told to isolate R is apparently about 1.5 in England.   It is hard to argue that the app is over reacting when despite the number of people getting pinged the number of infections is growing that fast.   

The actual problem is the app is doing what it is supposed to do but the government don't want to control Covid, they want to let it rip.  

And they are talking about starting to charge for tests.  They don't want people testing and isolating, they want it to spread and they'd like the monitoring to be less effective while their experiment is running so they don't need to explain the infection numbers.

Heavyside was a giant of engineering mathematics, up with the best ever. A self taught hobbyist, he attacked by the mathematics establishment for lack of rigour in amazingly useful engineering mathematics. He turned down a mathematics prize later in life as he never forgave them. He was deliberatly undermined on very useful engineering work by the plain nasty head of post office engineering. He seemed to be a very difficult man at the best of times and plain insulted people he regarded as fools.

His biography is well worth a read. An extended review here reads better than the rather unemotional wikiedia page. 

https://euro-math-soc.eu/review/forgotten-genius-oliver-heaviside-maverick-...

Sounds all very apt (no pun intended) given the conversations above, about peer review.

With infections high, the risk of receiving disruptive isolation orders high, various desirable things like festivals and clubs still closed pending decisions: this is probably the moment for the government to deploy its expertise in quietly but not too subtly twisting the electorate's arm to keep the jabs flowing. Alas they're too busy banging wedges into society ready for the next election to use their well developed media machinery and campaigning expertise for legitimate public health purposes.

It's not necessarily a problem the government or organisations with the appearance of government backing, can solve. Quite a bit of the reluctance is likely an unwillingness to engage with the state and documentation which can't be a new problem when you look at other issues like homelessness*, perhaps there's a creative role here for trusted charities and religious leadership. Likewise the young who aren't at much medical risk could perhaps be encouraged with some targeted campaigning more focused on the economic risk to them and the risk of another lost year of fun. Maybe add a carrot too as some US states experimented with, get a jab, get a lottery ticket... That'd have to be matched by very convenient jabs. It's annoying to jump through hoops for the reluctant and the ethics of bribes are a bit hmmmn but we need them jabbed so we need to think a bit further out of the 'use the NHS brand and hope' box. God knows what we do with the yummy-mummy wholefoods and crystals anti-vaxer sorts, dart guns**?

*because I'll get picked up on this by the professionally confused, I'm not saying the homeless are the immunity problem but they are a group where willingness to engage with the state is typically low but routinely worked around by those providing state-like services, there might be lessons to be learned.

**again, kidding

jk