New Covid rules for the UK coming into effect for...

[JohnD]

But we were told it would be doubling in 2-3 days, so should be eight times higher in a week.

you're concentrating on hospitalisation figures, which of course lag the infection rate by a few weeks.

It's already been postulated that with Omicron the rate of severe illness per number of cases will be lower, but the number of cases will be significantly higher.

edit

source: https://coronavirus.data.gov.uk/details/cases?areaType=overview&areaName=United Kingdom

I don't see the flattening-out you mention, especially bearing in mind that the last few days data is not all in yet.

 

[JohnD]

And here it is for the last month, to magnify the scale

(last few days incomplete)

The last three days is a steep rise, but it will take longer for the pattern to become clearer

 

[Justin Passing]

Justin Passing said: ↑
But we were told it would be doubling in 2-3 days, so should be eight times higher in a week.

you're concentrating on hospitalisation figures, which of course lag the infection rate by a few weeks.

We've had it long enough. It's valid still. It's a proportion, like 5% or something. If cases double, hospitalisations double. There's a time lag, but that doesn't affect the proportion.
See?

I don't see the flattening-out you mention, especially bearing in mind that the last few days data is not all in yet.

Signs of.
I gave a link - it's for London. The worst boroughs are certainly not going up as fast as they were and a couple have stalled.

Other areas of the UK will go up faster, (as eg Yorkshire has) because their feedstock is greater, from adjacent areas. If they just had a seed of one person, it would be slower.

There are many factors which influence the peaking. You can calculate a herd immunity figure from the R number easily, but there's more to it. Several are non-obvious unknowns though most of them don't make a huge difference.

 

[JohnD]

you mean, if 1,000 people are infected today, 50 of them go to hospital today?

No.

if 1,000 people are infected today, but 100 were infected 3 weeks ago, only 5 of them go to hospital today.

Because todays hospitalisations are not todays infections, but were infected a few weeks ago.

So at times of increase, hospitalisation will be a smaller proportion of cases.

 

[denso13]

"Robin Fransman, a well-known critic of the Dutch coronavirus policy and an outspoken skeptic of the Covid-19 vaccines, died on Tuesday from Covid-19."

https://nltimes.nl/2021/12/29/covid-19-kills-coronavirus-vaccine-skeptic-robin-fransman-53

 

[Deleted member 292770]

"Robin Fransman, a well-known critic of the Dutch coronavirus policy and an outspoken skeptic of the Covid-19 vaccines, died on Tuesday from Covid-19."

https://nltimes.nl/2021/12/29/covid-19-kills-coronavirus-vaccine-skeptic-robin-fransman-53

your excitement is palpable. what a disgusting person.

 

[IT Minion]

It's valid still. It's a proportion, like 5% or something. If cases double, hospitalisations double. There's a time lag, but that doesn't affect the proportion.
See?

The bit you're missing is that OMICRON is doubling every 2-3 days, but there was a steady baseline of Delta.

The Omicron hospitalisation case numbers have just started doubling to the point where they are clearly distinguishable from Delta.

So no, the hospitalisation rate won't be eight times what it is now in a week.

 

[Justin Passing]

No, there you go again - I'm not missing it, I'm referring to omicron cases, surely that's obvious. We are told omicron would double every 2-3 days. Delta has been fairly static numbers for a while.
Assuming we're testing a repsentative sample of the pop, we would see the doubling. We aren't testing as well, relatively, so we're guessing how widespread it is in the community.

you mean, if 1,000 people are infected today, 50 of them go to hospital today?

No.

Of course not.
If they were infected today, none would go to hospital today.

Of course each has a different delay, but each individual still has the same 5% (say) chance overall.
The number who go to hosp is still a multiple of the number infected. Some will take longer than others, so if you want an instantaneous figure you have to add some stats with a known distribution function. That would give a smeared-hump shaped graph which would come down again when the peaking function constraint took over. But the time integral would still be a multiple/divisor of the original cases.
The accumulating number who will go to hosp is an integral of the exponential function. For the number in hosp at one time you'd have to know the statistical function for how long people stay in, incorporate that into the admissions expression and integrate the lot. For large numbers/times you have to incorporate factors like reinfection and the reducing exponent as the feedstock gets used up, which is apart of the limiting function.

A common way to run the numbers is not to build one expression, but put stages of calculation for each function into Matlab which is good at chuntering through an array of functions. Or you can of course use a spreadsheet like excel. The incidental numbers go across, and you just total up the columns to integrate. Then you add your confidence limits on your functions and decide how well you want to cover the extremes of risk, for things like peak hospital occupancy.
Hence tents in car parks, even if they DO only cover an extra 4%. HMG will be able to claim they're not at fault, it'll be that unfotunate shortage of staff...

There are some good/simple pdfs on modeling, on the net.

For functions and integrals you'd need remember your school maths - A level or so, but a spreadsheet needs less knowledge to implement.

 

[IT Minion]

No, there you go again - I'm not missing it, I'm referring to omicron cases, surely that's obvious. We are told omicron would double every 2-3 days. Delta has been fairly static numbers for a while.
Assuming we're testing a repsentative sample of the pop, we would see the doubling. We aren't testing as well, relatively, so we're guessing how widespread it is in the community.



Of course not.
If they were infected today, none would go to hospital today.

Of course each has a different delay, but each individual still has the same 5% (say) chance overall.
The number who go to hosp is still a multiple of the number infected. Some will take longer than others, so if you want an instantaneous figure you have to add some stats with a known distribution function. That would give a smeared-hump shaped graph which would come down again when the peaking function constraint took over. But the time integral would still be a multiple/divisor of the original cases.
The accumulating number who will go to hosp is an integral of the exponential function. For the number in hosp at one time you'd have to know the statistical function for how long people stay in, incorporate that into the admissions expression and integrate the lot. For large numbers/times you have to incorporate factors like reinfection and the reducing exponent as the feedstock gets used up, which is apart of the limiting function.

A common way to run the numbers is not to build one expression, but put stages of calculation for each function into Matlab which is good at chuntering through an array of functions. Or you can of course use a spreadsheet like excel. The incidental numbers go across, and you just total up the columns to integrate. Then you add your confidence limits on your functions and decide how well you want to cover the extremes of risk, for things like peak hospital occupancy.
Hence tents in car parks, even if they DO only cover an extra 4%. HMG will be able to claim they're not at fault, it'll be that unfotunate shortage of staff...

There are some good/simple pdfs on modeling, on the net.

For functions and integrals you'd need remember your school maths - A level or so, but a spreadsheet needs less knowledge to implement.

Or just wait two weeks, because that rule of thumb has worked for the entirety of Covid.

 

[Justin Passing]

Or just wait two weeks, because that rule of thumb has worked for the entirety of Covid.

But we're looking forward.

 

[IT Minion]

But we're looking forward.

So you propose basing your complex model on...?

We don't have much good data on Omicron hospitalisation yet. It's pointless guessing if it's higher or lower, faster or slower yet. So we should continue to work based on Delta and see how it varies as that data comes in.

 

[Justin Passing]

Not my model and it's hardly complex. I already said, look at the numbers in SA. We're following earlier predictions which were made from there, and so we can see where we're going.
Ferguson said a couple of weeks ago we were looking at 5000 deaths a day from omicron.

Edit - thought I'd check that. Actually he said all sorts but 5000 was one of them:

 

[Justin Passing]

THE WHO on I thnk the 5th said there had been no deaths from omicron, and coetzee (SP) carried on quite a while after that saying they were only delta, zero omicron deaths.
It's gone up a bit now - - googling....

 

[IT Minion]

Not my model and it's hardly complex. I already said, look at the numbers in SA. We're following earlier predictions which were made from there, and so we can see where we're going.
Ferguson said a couple of weeks ago we were looking at 5000 deaths a day from omicron.

I don't think he did. I think you're misquoting him.

The screenshots you provide are a great example of shoddy reporting. Three reports on the same day, two from the same paper and they can't work out if it's 4,000 or 7,000.

Utterly ridiculous. How do you not look at that and question how they're getting such a spread?

 

[Deleted member 292770]

I don't think he did. I think you're misquoting him.

The screenshots you provide are a great example of shoddy reporting. Three reports on the same day, two from the same paper and they can't work out if it's 4,000 or 7,000.

Utterly ridiculous. How do you not look at that and question how they're getting such a spread?

Some people on here are so obsessed with regurgitating media bu!!sh!t with such conviction you’d swear it was their own .