Urban75 Home About Offline BrixtonBuzz Contact

Coronavirus in the UK - news, lockdown and discussion

I'd possibly be able to go much further with my comments if I knew what the structure and tech was like myself. But certainly its easy to imagine all sorts of ways to crudely work around this, even if it was only fit to be a temporary solution pending a more impressive update. Duplicate some tables so as not to risk messing up the originals (or use a dev copy of the system and data), add some fields, write and run some scripts to reprocess existing data as a result. Or depending on what sort of processing they do on raw data sources, if they are even in charge of that side of things within that system, there may be some relatively straightforward code changes that could get the job done.

Yes, I have found myself designing and maintaining databases for a living, by mistake, 20 years, man and boy etc. Something can always be kludged, especially if it’s output only (like the dashboard).

My blindness is that it’s not clear to me how one could structure the data in the first place such that adding an A/not-A/all filter on a binary* attribute wouldn’t be close to trivial. In fact in many (perfectly sensible) structures it would be harder to exclude 2nd and subsequent infections. But maybe I’m lacking in imagination.

* even if it’s an ordinal or first-count it’s still going to be A/not-A at the output I guess.
 
Yes, I have found myself designing and maintaining databases for a living, by mistake, 20 years, man and boy etc. Something can always be kludged, especially if it’s output only (like the dashboard).

My blindness is that it’s not clear to me how one could structure the data in the first place such that adding an A/not-A/all filter on a binary* attribute wouldn’t be close to trivial. In fact in many (perfectly sensible) structures it would be harder to exclude 2nd and subsequent infections. But maybe I’m lacking in imagination.

* even if it’s an ordinal or first-count it’s still going to be A/not-A at the output I guess.
Yeah. I need to know more about source data and how it is processed.

We know that at one stage the data got truncated because they were using a spreadsheet, so thats not a promising sign of how things used to be. Although maybe that was for deaths, I cant remember.

If they've got raw data thats linked to individuals then they need to change their scripts that currently only allow one individual to show up once, the first time, in the simplified output from that processing. How this is then represented in tables that are exposed to the public via graphs and data downloads is another matter, more than one way to do it, most of which are probably trivial in some ways and slightly hellish to develop in others. I dont even know what they have in mind for how they present this data, probably via some specific graphs, but these reinfections should also be a part of the main headline totals by date of reporting and date of test specimen.

Wales doesnt have this limitation, adn their resulting data is used on the main UK dashboard. They decided to define a single episode of someone catching covid by using a 6 week window. I've been trying to find out how long ago they made this change. I havent succeeded, maybe they did this since the start or quite early on, but I at least discovered that this is described in a document that says it hasnt been updated since August 2020:

Individuals may be tested more than once for COVID-19 for numerous reasons. A testing episode is a six-week period starting from the date of the first sample taken from the patient. Individuals who are tested multiple times during a six-week period are only counted once during that period.

If any of the test results for the individual are positive then that is the result which is presented. If an individual tests positive more than once during the six-week period then this is still recorded as only one new case.

Any tests which occur more than six weeks after the initial test will trigger a new testing episode.


I do much prefer the user interface of the main UK dashboard and the range of things it shows, compared to the tableau ones the likes of the Welsh government use ( https://public.tableau.com/app/prof.../RapidCOVID-19virology-Public/Headlinesummary )
 
Last edited:
I'm really glad, now, that the once-widely-spread local rumour that she was going to take over and manage the longtime-defunct former Adan & Eve further up the High Street nearer the station, never came about.

At the moment the place, which still has the pub sign saying 'The Last Resort' (briefly excellent bar ... ), has become something under a different name, 'The Hippo Bar'

Which sounds really fucking crap :hmm:

But a mate of ours, who may once have 'known some people' more 'nvolved' ;) with weed dealing many years ago ;), has taken over, and he's started to be pretty cool at sourcing good beer. For now, just on electric fizz-machine pumps, but the beers were far from bad last week .... Tiny Rebel was included ...

And once trade ends up recovering, he plans to revive the actual cellar and handpumps with class Welsh ales such as Greytrees and Gower, plus even better imports ......

For now though, he's closed the place :( -- insufficient staff to do table service, and he's a one-man boss ....

Still, he reckons Drakeford will cease or at least reduce the current Welsh rules by early-to-mid-February.

That, for the time being, seems fairly plausible in Wales IMO :)

(Apologies for derail .... :oops: ;) )
 
Last edited:
“We’re sorry you feel your hospitals are overwhelmed”
That side of todays press conference was hugely depressing.

The press werent exactly convinced by it either, judging by many of the questions.

I'd be going far more nuts about this if I hadnt been prepared for it for so long. That doesnt help much, but at least it wasnt a shock. I tiwll probably take me longer to get over some of the attitudes expressed here in recent weeks.
 
Yes, I have found myself designing and maintaining databases for a living, by mistake, 20 years, man and boy etc. Something can always be kludged, especially if it’s output only (like the dashboard).

My blindness is that it’s not clear to me how one could structure the data in the first place such that adding an A/not-A/all filter on a binary* attribute wouldn’t be close to trivial. In fact in many (perfectly sensible) structures it would be harder to exclude 2nd and subsequent infections. But maybe I’m lacking in imagination.

* even if it’s an ordinal or first-count it’s still going to be A/not-A at the output I guess.
I'd guess that either the data haven't been captured consistently, or that they have, but that publishing an analysis would reveal some other fuck up.
 
They've got some forms of the data because they do reinfection reports, but only monthly in recent times. And those reports will show a large leap in the Omicron wave. I'll point one out next time its updated.
 
For the first time in my life I found myself thinking 'I hope I don't need the hospital any time soon.' I mean obviously we all hope that but for reasons of health. I've never thought it for fear of having to wait 12 hours in a packed waiting room or queuing ambulance, if I'm lucky, to get seen by over stressed and exhausted staff.

I was listening to nursing staff phoning the radio yesterday and every one that called in said that things have either been overwhelmed or are on the verge of being overwhelmed.
 
It would be catastrophic to release information that would put massive numbers of people off attending hospital that need care as they will ultimately end up there anyway in a far worse condition putting way more pressure on nursing teams than if they attended when their condition was manageable. Straight to ICU don't pass go.
 
Yes, I have found myself designing and maintaining databases for a living, by mistake, 20 years, man and boy etc. Something can always be kludged, especially if it’s output only (like the dashboard).

My blindness is that it’s not clear to me how one could structure the data in the first place such that adding an A/not-A/all filter on a binary* attribute wouldn’t be close to trivial. In fact in many (perfectly sensible) structures it would be harder to exclude 2nd and subsequent infections. But maybe I’m lacking in imagination.

* even if it’s an ordinal or first-count it’s still going to be A/not-A at the output I guess.
I did databases for a living, and my suspicion is that they've probably - as so many do - designed the database appallingly, without any appropriate normalisation that would enable subtle changes to data structure to be made without horrific restructuring of existing data. Towards the end, most of my DB work was trying to untangle the shitty mess that previous design assumptions had led to. And that wasn't in the public sector, which I suspect is 10x worse.
 
Test kits Shrinking like Mars bars
 

Attachments

  • 270813465_10160466005584644_5435182210213662668_n.jpg
    270813465_10160466005584644_5435182210213662668_n.jpg
    95.6 KB · Views: 36
Looks like a good way to enable the royal mail etc to deliver more of them to me. And/or they ran out of larger boxes and managed to source some smaller ones.
 
Looks like a good way to enable the royal mail etc to deliver more of them to me. And/or they ran out of larger boxes and managed to source some smaller ones.
There are less tests in the kit and they are for NHS staff to test before going on shift so many wtfs shared
 
It would be catastrophic to release information that would put massive numbers of people off attending hospital that need care as they will ultimately end up there anyway in a far worse condition putting way more pressure on nursing teams than if they attended when their condition was manageable. Straight to ICU don't pass go.
That happened a lot already, especially in the first wave. It was down to a few factors - plenty of people figures out that hospital spread was a thing, but also the central 'stay home, protect the NHS' message caused some people to go to extremes, deciding not to seek medical intervention in order to reduce the NHS burden. Plus some people get very scared at needing to go to hospital anyway, and an excuse to ignore it and hope they recover anyway had some appeal. Plus the authorities didnt put a brilliant system for evaluating people remotely in place, the telephone triage wasnt very good, and things were deliberately put in place to reduce demand during the peak. They may have been expecting the first wave to be even larger than it was, so they went overboard with demand destruction. People were also not given much information about what some of the extreme danger signs were in regards catastrophic complications of having blood oxygen levels that were too low, and some people that did seek late intervention were still not admitted anyway.

That awful picture is rather visible in certain first wave data, most obviously the deaths from all causes during the period, which really stands out for the first wave in ways that were not the same in subsequent waves, not even the awful 2nd wave.
 
That would be a bit fucking daft
Its not necessarily a bad way to adjust a system to cope with supply/demand issues, but only so long as a continual dribble of the modified supply can reach the right people at the right time not to leave a different sort of gap in their ability to test.
 
Fair enough.

Anyway it wont surprise anyone to know that the way I would ideally handle such issues is by not allowing a wave to go beyond a certain size in the first place.
 
It would be catastrophic to release information that would put massive numbers of people off attending hospital that need care as they will ultimately end up there anyway in a far worse condition putting way more pressure on nursing teams than if they attended when their condition was manageable. Straight to ICU don't pass go.
By the way, Wales had no problem putting hospital covid infections on their dashboard.

Screenshot 2022-01-05 at 13.16.jpg

Some data for England is released by NHS England too, but in convoluted form and not via graphs on the main UK dashboard.
 
Back
Top Bottom