| 1 in 5 has had covid according to a new study on 20:40 - Jan 10 with 1956 views | J2BLUE |
No chance IMO |  |
|  |
| 1 in 5 has had covid according to a new study on 20:45 - Jan 10 with 1934 views | bluelagos |
The methodology is the computer models they use. It is based on the theory that you have a population (say P) which is split between infected, previously infected and still to be infected. So as the no. of previously infected rises, the R rate will drop as there are less new people to infect. That is basically what herd immunity is. You can only pass it on to those still to be infected.
So the figures in East London being around 50% previously infected then the R will drop as all those wont (theoretically) get it twice. In fact if the R is around 1 in East London then the 1 in 6 currently infected will pass it on to another 16% so they will be reaching herd immunity levels relatively soon in some parts of East London (sadly not elsewhere)
The other good thing is that the 2 models are aligned to thats a comfort blanket they are accurate.
The statement that infection rates are between 4 and 8 times offcial figures seems quite wide. I guess the lack of reliable community testing means there is still some uncertainty there.
Edit: Just googled it and its called an S-I-R model which stands for susceptible, infected and removed.
Edit2: In case anyone is in doubt, I am not advocating herd immunuty through allowing the disease to spread. Just clumsily explaining the way the models work. [Post edited 11 Jan 2021 7:16]
|  |
| |
| 1 in 5 has had covid according to a new study on 20:46 - Jan 10 with 1922 views | bluelagos |
| 1 in 5 has had covid according to a new study on 20:40 - Jan 10 by J2BLUE |
No chance IMO |
Why not? The accuracy of these models has been born out by the recent rises in cases that they forecast a number of weeks ago. | |
| |
| 1 in 5 has had covid according to a new study on 20:50 - Jan 10 with 1893 views | WD19 |
6% had had it nationally after the first wave.
So it seems a bit toppy, but not beyond the realms of possibility.
Imperial are launching a further survey to measure it this week. |  | | |
| 1 in 5 has had covid according to a new study on 20:51 - Jan 10 with 1884 views | J2BLUE |
| 1 in 5 has had covid according to a new study on 20:46 - Jan 10 by bluelagos |
Why not? The accuracy of these models has been born out by the recent rises in cases that they forecast a number of weeks ago. |
Maybe it's from my sheltered Ipswich viewpoint but it just doesn't fit from what i've seen. Unless there have been a huge number of asymptomatic cases.
There's a chance of course that in Ipswich it's 1 in 20 and in London/other hot spots it's vastly more but i'm not convinced. We've seen these kind of estimates before. | |
| |
| 1 in 5 has had covid according to a new study on 20:53 - Jan 10 with 1864 views | bluelagos |
| 1 in 5 has had covid according to a new study on 20:45 - Jan 10 by bluelagos |
The methodology is the computer models they use. It is based on the theory that you have a population (say P) which is split between infected, previously infected and still to be infected. So as the no. of previously infected rises, the R rate will drop as there are less new people to infect. That is basically what herd immunity is. You can only pass it on to those still to be infected.
So the figures in East London being around 50% previously infected then the R will drop as all those wont (theoretically) get it twice. In fact if the R is around 1 in East London then the 1 in 6 currently infected will pass it on to another 16% so they will be reaching herd immunity levels relatively soon in some parts of East London (sadly not elsewhere)
The other good thing is that the 2 models are aligned to thats a comfort blanket they are accurate.
The statement that infection rates are between 4 and 8 times offcial figures seems quite wide. I guess the lack of reliable community testing means there is still some uncertainty there.
Edit: Just googled it and its called an S-I-R model which stands for susceptible, infected and removed.
Edit2: In case anyone is in doubt, I am not advocating herd immunuty through allowing the disease to spread. Just clumsily explaining the way the models work. [Post edited 11 Jan 2021 7:16]
|
Just to add that vaccinations do a few things.
1. If they offer protection they move people from the still be infected part to the previously infected part of the model bringing herd immunity closer.
2. Those who are infected are less likely to die.
3. Those infected are less contagious and so R drops.
All of these will be modelled as best they can, albeit a lot of it is best guesses given the number of unknowns.
***All based on when I studied this as part of my degree 30 years ago. Am sure some things will have changed. And I have never done any of this in the workplace. | |
| |
| 1 in 5 has had covid according to a new study on 20:57 - Jan 10 with 1852 views | bluelagos |
| 1 in 5 has had covid according to a new study on 20:51 - Jan 10 by J2BLUE |
Maybe it's from my sheltered Ipswich viewpoint but it just doesn't fit from what i've seen. Unless there have been a huge number of asymptomatic cases.
There's a chance of course that in Ipswich it's 1 in 20 and in London/other hot spots it's vastly more but i'm not convinced. We've seen these kind of estimates before. |
Ok so one way to look at it is we originally estimated 1% of people die if we do nothing. I.e. 80% of 65m get infected and 1% die is 500k people.
Well given we are now approaching 100k deaths ( by end Jan?) that suggests around 1/5 of the population have been infected... | |
| |
| 1 in 5 has had covid according to a new study on 20:58 - Jan 10 with 1838 views | Nthsuffolkblue |
I can't see anywhere what they base their figures on.
Even if true, the 1 in 5 rising to 1 in 2 in London still leaves the majority (80%) of the population to have not had any chance to develop an immunity.
Rising to 1 in 2 in London would suggest half the population of London is still as vulnerable as before. It also means that number is higher than 80% in other parts of the country to allow for that skewing.
The possibility that infection does not equal immunity and that asymptomatic sufferers (or even symptomatic ones) may be able to catch it again is another factor to consider.
I think the only way we will get herd immunity without loss of life in the many hundreds of thousands is by mass vaccination. |  |
| |
Login to get fewer ads
| 1 in 5 has had covid according to a new study on 21:03 - Jan 10 with 1802 views | bluelagos |
| 1 in 5 has had covid according to a new study on 20:58 - Jan 10 by Nthsuffolkblue |
I can't see anywhere what they base their figures on.
Even if true, the 1 in 5 rising to 1 in 2 in London still leaves the majority (80%) of the population to have not had any chance to develop an immunity.
Rising to 1 in 2 in London would suggest half the population of London is still as vulnerable as before. It also means that number is higher than 80% in other parts of the country to allow for that skewing.
The possibility that infection does not equal immunity and that asymptomatic sufferers (or even symptomatic ones) may be able to catch it again is another factor to consider.
I think the only way we will get herd immunity without loss of life in the many hundreds of thousands is by mass vaccination. |
Absolutely. Just because parts of East London are getting close to herd immunity is no argument that we should relax anything.
We will acheive herd immunity but it is far better to do so through vaccinations than allowing the virus to kill 100s of thousands more people.
It really is a race to protect people asap. Which is what we are doing tbf. | |
| |
| 1 in 5 has had covid according to a new study on 21:05 - Jan 10 with 1793 views | Darth_Koont |
| 1 in 5 has had covid according to a new study on 20:45 - Jan 10 by bluelagos |
The methodology is the computer models they use. It is based on the theory that you have a population (say P) which is split between infected, previously infected and still to be infected. So as the no. of previously infected rises, the R rate will drop as there are less new people to infect. That is basically what herd immunity is. You can only pass it on to those still to be infected.
So the figures in East London being around 50% previously infected then the R will drop as all those wont (theoretically) get it twice. In fact if the R is around 1 in East London then the 1 in 6 currently infected will pass it on to another 16% so they will be reaching herd immunity levels relatively soon in some parts of East London (sadly not elsewhere)
The other good thing is that the 2 models are aligned to thats a comfort blanket they are accurate.
The statement that infection rates are between 4 and 8 times offcial figures seems quite wide. I guess the lack of reliable community testing means there is still some uncertainty there.
Edit: Just googled it and its called an S-I-R model which stands for susceptible, infected and removed.
Edit2: In case anyone is in doubt, I am not advocating herd immunuty through allowing the disease to spread. Just clumsily explaining the way the models work. [Post edited 11 Jan 2021 7:16]
|
Although I’m still not sure how accurately we’ve modelled transmission or if we can yet. Just taking a look at similar sized countries like UK, France, Germany, Spain and Italy and the infection rate has exploded or been relatively benign – and seemingly irrespective of particular measures. It seems as if there’s an asymptomatic spread and a seriously symptomatic spread too, whether they’re the same or overlapping.
It will be fascinating to see how exactly Covid has spread (and which variants) once all the data can be collected.
I think that’s why the vaccine is so important. Anything but zero Covid approaches can’t be controlled. |  |
| |
| 1 in 5 has had covid according to a new study on 21:13 - Jan 10 with 1742 views | bluelagos |
| 1 in 5 has had covid according to a new study on 20:58 - Jan 10 by Nthsuffolkblue |
I can't see anywhere what they base their figures on.
Even if true, the 1 in 5 rising to 1 in 2 in London still leaves the majority (80%) of the population to have not had any chance to develop an immunity.
Rising to 1 in 2 in London would suggest half the population of London is still as vulnerable as before. It also means that number is higher than 80% in other parts of the country to allow for that skewing.
The possibility that infection does not equal immunity and that asymptomatic sufferers (or even symptomatic ones) may be able to catch it again is another factor to consider.
I think the only way we will get herd immunity without loss of life in the many hundreds of thousands is by mass vaccination. |
Ref your first comment.
What they do is build a forecasting model with a whole range of assumptions around the infection rates, transmission rates and thus forecast how many people have previosuly been infected as well as future infections.
They will test it on the previous data (comparing it to all the info they have) and then go and refine it. So next week they are maybe forecasting 5k deaths, if they get 4k they will refine some of their assumptions to see if they can reforecast 4k. It could be the transmission rate, the overall infections or the death rate of those ill.
None of it is exact, but after 9 months of data they will be getting much more reliable as models. And that the 2 main models have similar results in another indicator they are pretty accurate.
And as I said to J2, most of their warnings in early Dec have been shown to be spot on, indicating they are fairly robust models. | |
| |
| 1 in 5 has had covid according to a new study on 21:14 - Jan 10 with 1732 views | J2BLUE |
| 1 in 5 has had covid according to a new study on 20:57 - Jan 10 by bluelagos |
Ok so one way to look at it is we originally estimated 1% of people die if we do nothing. I.e. 80% of 65m get infected and 1% die is 500k people.
Well given we are now approaching 100k deaths ( by end Jan?) that suggests around 1/5 of the population have been infected... |
I have no idea if this logic stands up to scientific analysis.
For the whole country it may well be true and i'd find it easier to believe if they said 1 in 20 in Ipswich.
If 1/5 really have been infected can't we test people and prioritise vaccines like Blair suggested? | |
| |
| 1 in 5 has had covid according to a new study on 21:19 - Jan 10 with 1714 views | bluelagos |
| 1 in 5 has had covid according to a new study on 21:05 - Jan 10 by Darth_Koont |
Although I’m still not sure how accurately we’ve modelled transmission or if we can yet. Just taking a look at similar sized countries like UK, France, Germany, Spain and Italy and the infection rate has exploded or been relatively benign – and seemingly irrespective of particular measures. It seems as if there’s an asymptomatic spread and a seriously symptomatic spread too, whether they’re the same or overlapping.
It will be fascinating to see how exactly Covid has spread (and which variants) once all the data can be collected.
I think that’s why the vaccine is so important. Anything but zero Covid approaches can’t be controlled. |
There will always be an element of guess work. So they are never 100% accurate. That they stated they think the current true rates of infections are between 4 and 8 times the confirmed tests shows there is still a fair bit of uncertainty in the modelling.
But they are the being done by the best mathmatical modellers of diseases in the country. These guys have spent years developing and refining what they do and have amazing computing power to simulate the disease and all the factors around it. | |
| |
| 1 in 5 has had covid according to a new study on 21:23 - Jan 10 with 1700 views | Darth_Koont |
| 1 in 5 has had covid according to a new study on 21:19 - Jan 10 by bluelagos |
There will always be an element of guess work. So they are never 100% accurate. That they stated they think the current true rates of infections are between 4 and 8 times the confirmed tests shows there is still a fair bit of uncertainty in the modelling.
But they are the being done by the best mathmatical modellers of diseases in the country. These guys have spent years developing and refining what they do and have amazing computing power to simulate the disease and all the factors around it. |
Absolutely. If anyone is going to do the modeling it’s them.
Not sure the data quality is there though. Seems like we’re still playing catch up. | |
| |
| 1 in 5 has had covid according to a new study on 21:26 - Jan 10 with 1692 views | bluelagos |
| 1 in 5 has had covid according to a new study on 21:14 - Jan 10 by J2BLUE |
I have no idea if this logic stands up to scientific analysis.
For the whole country it may well be true and i'd find it easier to believe if they said 1 in 20 in Ipswich.
If 1/5 really have been infected can't we test people and prioritise vaccines like Blair suggested? |
We can, and thats what we call community testing. Think Rjorry and a few others have been having regular tests and that will inform the modellers of the current rates of infection.
Whether they are doing the antibody testing have no idea. One problem though is testing only seems to pick up 60% of those infected (according to Marr this morning quoting the results from mass testing in Liverpool) | |
| |
| 1 in 5 has had covid according to a new study on 21:30 - Jan 10 with 1676 views | bluelagos |
| 1 in 5 has had covid according to a new study on 21:23 - Jan 10 by Darth_Koont |
Absolutely. If anyone is going to do the modeling it’s them.
Not sure the data quality is there though. Seems like we’re still playing catch up. |
We are playing catch up, but that is unavoidable. The models will always be inaccurate, especially at the start of a pandemic.
But 9 months of data will make them a lot more accurate. | |
| |
| 1 in 5 has had covid according to a new study on 21:41 - Jan 10 with 1637 views | clive_baker |
Sounds plausible to me. Would mean the virus has a mortality rate of c. 0.6% or 1 in 167 which feels about right. |  |
| |
| 1 in 5 has had covid according to a new study on 22:07 - Jan 10 with 1573 views | Darth_Koont |
| 1 in 5 has had covid according to a new study on 21:30 - Jan 10 by bluelagos |
We are playing catch up, but that is unavoidable. The models will always be inaccurate, especially at the start of a pandemic.
But 9 months of data will make them a lot more accurate. |
Yes, I just haven’t seen much of the K value and that for example 80% of transmissions come from 10% of infected individuals.
That suggests a necessary leeway in the modelling or at least that you overcompensate on the measures when you see an increase.
This is really not a criticism of the science or the models, I think the problem is that we have a novel virus going head to head with established political ego. | |
| |
| 1 in 5 has had covid according to a new study on 22:22 - Jan 10 with 1513 views | Nthsuffolkblue |
| 1 in 5 has had covid according to a new study on 21:03 - Jan 10 by bluelagos |
Absolutely. Just because parts of East London are getting close to herd immunity is no argument that we should relax anything.
We will acheive herd immunity but it is far better to do so through vaccinations than allowing the virus to kill 100s of thousands more people.
It really is a race to protect people asap. Which is what we are doing tbf. |
Thanks, yes I agree and we do appear to be doing the right thing (pretty much) now, albeit having not done so for too long and only dong so to a degree from the pressure from teaching unions. Indeed there is a reasonable argument that because it was left so long, we really need tougher restrictions right now. There was a lot of traffic on the road Friday afternoon. | |
| |
| 1 in 5 has had covid according to a new study on 06:32 - Jan 11 with 1341 views | dickie |
How possible is herd immunity if it seems that people can catch Covid more than once? |  | | |
| 1 in 5 has had covid according to a new study on 06:54 - Jan 11 with 1320 views | gordon |
There won't be herd immunity, just seasonal outbreaks each winter, the severity of which will depend on what strain is dominant and how well matched the vaccine is. No-one should ever have mentioned herd immunity in relation to COVID, because it just isn't a relevant concept. |  | | |
| 1 in 5 has had covid according to a new study on 07:12 - Jan 11 with 1299 views | dickie |
| 1 in 5 has had covid according to a new study on 06:54 - Jan 11 by gordon |
There won't be herd immunity, just seasonal outbreaks each winter, the severity of which will depend on what strain is dominant and how well matched the vaccine is. No-one should ever have mentioned herd immunity in relation to COVID, because it just isn't a relevant concept. |
Should they have called it "Operation Sacrifice The Oldies" back when the government were talking about herd immunity? | | | |
| 1 in 5 has had covid according to a new study on 08:02 - Jan 11 with 1253 views | bluelagos |
Going to rack my brains and try and get technical. Feel frèe to ignore if the concepts worry you.
All based on what we learned at Uni 30 years ago but I think the basic premise is largely unchanged.
So think of a local village with 1000 people. To build a model of how the disease spreads you need to estimate a number of things. How many have the disease now (Infected) how many have had it before (recovered) and how many have yet to have it (susceptible), how fast it transmits and to how many (R rate) and lastly, how many of those who get it, will sadly die.
The modelling is all about estimating these numbers as you dont have the ability to test every person, every day. You use whatever info you have to fine your estimates and remodel. You hope to then inform the policy makers to make informed decisions.
So our village of 1000 people. You hear about the disease near by but so far no one has fallen ill. At this stage there is uncertainty. Maybe 5 people have it but are not symptomatic yet? Maybe some are ill but just hate going to the doctor? Maybe some and ill with other things and you dont notice? That means you make a rough guess.
So I as a modeller say given no one is ill I think between 0 and 10 have the illness and my best estimate is 3. I think it unlikely anyone is now recovered as the illness is just starting to I estimate 0 are recovered.
A week later, we get 4 people present to hospital. Now we test them and 3 are ill with the disease. So we know 3 are definitely infected. But we also know, as before, others maybe too. But how many? It is still guess work but we can say we think the infectious are in the range 3 to 30 and our best estimate is say 7.
Over a few days we can now estimate our R value, the rate at which the disease is spreading. Again it's a best guess but is based on everything we know at this time.
Go forward a few weeks and we can see people coming into the doctors are growing at 10% every week and we sadly now have people dieing. So as a modeller I can use all the info I have to model the course of the disease if we do nothing. I do this by forecasting how ths disease spreads, how many people are still to get ill (susceptible) and if/when they will get ill.
So for 1000 people I know now 25 have been to the doctors and been tested but I also need to estimate how many were ill but didnt go to the doctors.
I do this as best I can knowing it will have a level of uncertainty and am clear about that knowing some people want more accurate modelling.
So lets say I am now a few months later and the disease is further along. Now the number of people falling ill is say 5%, 1 in 20 are ill. This is higher than I early modelled/prediicted.
There are number of possibilities. Either the R has jumped higher (say a new variant) or maybe the spread was higher than I realised. Maybe I need to refine my estimate going all the way back. I can test my new assumptions and refine my model to see if any different assumptions would better predict what happened.
There is still uncertainty in my modelling, but is getting better and more accurate.
Now this is getting difficult right? And this is for 1 village, multiple the village results by 65000 other sets of 1000 people and we have the UK, so it clearly very complex.
So modelling how a disease grows through a village is doable. That is basically what we did at Uni as final year student (part of my degree was mathematical modellng) we modelled how the plague went though Eyam. All the villagers closed themselves off and so the village records of deaths can be used to model how a disease spreads.
This model (S-I-R) is what the guys are using, only way more complex. These guys are having to do something way more complex than I described because they have more villages/towns to cover. They have to do it as they go, they have to throw in lockdowns, compliance, international travellers, mutations etc.
On the plus side they now have huge amounts of data. All the testing, all the rates of deaths, months worth of ups and downs to model the disease and its infectious rates, susceptibles and recovereds.
Still lots of uncertainty am sure, but these guys are the best there is. They are doing all they can and their models are getting better. We can choose to listen to them or to some bloke on facebook who read something on the internet....
The last sentence, it is really is that simple.
Edit: Been pointed out the modelling in the OP is not "S-I-R" modelling - so worth pointing that out. [Post edited 11 Jan 2021 8:42]
| |
| |
| 1 in 5 has had covid according to a new study on 08:16 - Jan 11 with 1232 views | gordon |
| 1 in 5 has had covid according to a new study on 08:02 - Jan 11 by bluelagos |
Going to rack my brains and try and get technical. Feel frèe to ignore if the concepts worry you.
All based on what we learned at Uni 30 years ago but I think the basic premise is largely unchanged.
So think of a local village with 1000 people. To build a model of how the disease spreads you need to estimate a number of things. How many have the disease now (Infected) how many have had it before (recovered) and how many have yet to have it (susceptible), how fast it transmits and to how many (R rate) and lastly, how many of those who get it, will sadly die.
The modelling is all about estimating these numbers as you dont have the ability to test every person, every day. You use whatever info you have to fine your estimates and remodel. You hope to then inform the policy makers to make informed decisions.
So our village of 1000 people. You hear about the disease near by but so far no one has fallen ill. At this stage there is uncertainty. Maybe 5 people have it but are not symptomatic yet? Maybe some are ill but just hate going to the doctor? Maybe some and ill with other things and you dont notice? That means you make a rough guess.
So I as a modeller say given no one is ill I think between 0 and 10 have the illness and my best estimate is 3. I think it unlikely anyone is now recovered as the illness is just starting to I estimate 0 are recovered.
A week later, we get 4 people present to hospital. Now we test them and 3 are ill with the disease. So we know 3 are definitely infected. But we also know, as before, others maybe too. But how many? It is still guess work but we can say we think the infectious are in the range 3 to 30 and our best estimate is say 7.
Over a few days we can now estimate our R value, the rate at which the disease is spreading. Again it's a best guess but is based on everything we know at this time.
Go forward a few weeks and we can see people coming into the doctors are growing at 10% every week and we sadly now have people dieing. So as a modeller I can use all the info I have to model the course of the disease if we do nothing. I do this by forecasting how ths disease spreads, how many people are still to get ill (susceptible) and if/when they will get ill.
So for 1000 people I know now 25 have been to the doctors and been tested but I also need to estimate how many were ill but didnt go to the doctors.
I do this as best I can knowing it will have a level of uncertainty and am clear about that knowing some people want more accurate modelling.
So lets say I am now a few months later and the disease is further along. Now the number of people falling ill is say 5%, 1 in 20 are ill. This is higher than I early modelled/prediicted.
There are number of possibilities. Either the R has jumped higher (say a new variant) or maybe the spread was higher than I realised. Maybe I need to refine my estimate going all the way back. I can test my new assumptions and refine my model to see if any different assumptions would better predict what happened.
There is still uncertainty in my modelling, but is getting better and more accurate.
Now this is getting difficult right? And this is for 1 village, multiple the village results by 65000 other sets of 1000 people and we have the UK, so it clearly very complex.
So modelling how a disease grows through a village is doable. That is basically what we did at Uni as final year student (part of my degree was mathematical modellng) we modelled how the plague went though Eyam. All the villagers closed themselves off and so the village records of deaths can be used to model how a disease spreads.
This model (S-I-R) is what the guys are using, only way more complex. These guys are having to do something way more complex than I described because they have more villages/towns to cover. They have to do it as they go, they have to throw in lockdowns, compliance, international travellers, mutations etc.
On the plus side they now have huge amounts of data. All the testing, all the rates of deaths, months worth of ups and downs to model the disease and its infectious rates, susceptibles and recovereds.
Still lots of uncertainty am sure, but these guys are the best there is. They are doing all they can and their models are getting better. We can choose to listen to them or to some bloke on facebook who read something on the internet....
The last sentence, it is really is that simple.
Edit: Been pointed out the modelling in the OP is not "S-I-R" modelling - so worth pointing that out. [Post edited 11 Jan 2021 8:42]
|
The analysis in the OP isn't a SIR model, they used a fatality rate and number of deaths to estimate total infections. It's quite a simplistic way of doing it (particularly when there's a high proportion of asymptomatic infections) but won't be a million miles away. | | | |
| 1 in 5 has had covid according to a new study on 08:20 - Jan 11 with 1225 views | bluelagos |
| 1 in 5 has had covid according to a new study on 08:16 - Jan 11 by gordon |
The analysis in the OP isn't a SIR model, they used a fatality rate and number of deaths to estimate total infections. It's quite a simplistic way of doing it (particularly when there's a high proportion of asymptomatic infections) but won't be a million miles away. |
Ah, well I'll leave it up (And edit) as I know some people may find it handy to understand how the S-I-R models work. | |
| |
| |
By continuing to use the site, you agree to our use of cookies and to abide by our Terms and Conditions.
