I have no idea if the idea in this video is really deliverable, or not, although Rwanda is apparently using it, as has China.
But the idea seems powerful, and with appropriate logistical management it seems deliverable.
It's worth 15 minutes of your time for the sheer intellectual curiosity of it, frankly.
Take a look if you have a spare moment.
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
I usually agree with most of your opinions and viewpoints.
I have to say that this is one of the most confusing videos I have ever read.
I do not even begin to understand anything that this man is talking about. he talks to fast and assumes that you have understood and are capable of accessing to a level of comprehension that most people do not aspire to.
Can you please ensure that in future videos which are imported to your site are understandable to the ordinary citizen. Thank you very much
Noted
I enjoyed it
And I admit, this is my blog…
If people found the video confusing — have a listen to R4’s ‘More of Less’ from about 2 weeks ago. They reviewed the same approach. It is being used in Israel. A more basic mass sampling approach first used in WW2 to help identify US troops with STD’s.
I’ve used this technique (although I didn’t know it had a maths background) to fault find on large control panels. It works brilliantly and saves a lot of time.
I’ve sent the URL to my MP. As she’s a Brexit Tory, it’ll either get forwarded to the right people or totally ignored. I also included a link to George Monbiot’s article in today’s Guardian asking for her comments. I won’t hold my breath.
Robin – you beat me to it – I was going to send Monbiot’s article to our MP and asked her to dispute any part of it.
What a fascinating idea. Interesting that he mentions that China is using a “simple” pooling when they mass test millions. Also the technique is used in Rwanda. The fear would be that the combinatorial phase could be mishandled with disastrous effects – I would not trust the present UK testing system to cope reliably with the necessary level of complexity. But hey – bring in the mathematicians to help – at least they will be able to think logically unlike ….!
Heard about this on Radio 4 Today, a few days ago.. (sometimes there are useful things on there..) it said to split each persons sample into six samples and mix those with samples form others tested. By getting the maths right, you could identify the individuals that were covid positive. At the same time you significantly reduced the number of lab tests required.
So that could work for e.g. NHS and care home staff, where you could increase testing without adding to pressure on labs.
I liked it and it got me thinking how little we were doing in UK on testing generally – as in using maths to help identify e.g. superspreaders and highlighting where to look for the problems.
All we seem to do is test people as they have ‘need’ or are able to get a test and ignore what the numbers could tell us if it was organised better using a mathematical approach.
Perhaps it would tell us what is happening in some of the hotspots.
Couple that with local tracing and a number of useful patterns might emerge.
Meanwhile Serco shareholders will collect their dividend….
Theoretically you cannot argue with maths. But my faith in humans actually applying it correctly, public or private, is very low. One mistake would be disastrous.
This works as far as it goes. And simple pooling is already being done in places eg by some universities. The great thing about maths is there is no opinion involved. Either it has been proved, or it hasn’t. If proven, it works.
The problem this won’t help with is the one we have in Australia, which is two few people going for tests, because there’s so little virus. For that we need simpler, faster and less invasive ways of administering tests, especially self-administered. At that point the numbers might go up enough that this maths would make a difference.
Neat. The concept can be expressed in one sentence. “If the prevalence of a disease is low enough, you need many fewer tests if you combine samples from several people together and test as groups rather than individually, and then retest only the individuals in groups with a positive result.”
As a result, in the low prevalence regime, the same number of tests can be used to screen a much larger number of people (by a multiple of several) by grouping their tests together.
No doubt there is some equally clever maths that explains how to adapt the theory to deal with a real test, with false positives and false negatives.
No, it will not help.
Note that this system needs a fully working test trace isolate system to follow up the positive cases promptly and thoroughly — something the presenter only mentions briefly.
Despite vast expenditure the UK Government has not managed to get a reliable test trace isolate system working properly in England.
Ramping up the numbers of people tested is utterly useless without that proper follow up. There is no point in considering the use of any variant of mass or pooled test systems until the our basic public health systems have been effectively restored. It would just be another project for this incompetent corrupt Tory government to spaff money on.
I suggest you access the video, audio read the transcripts on the WHO website https://www.who.int/emergencies/diseases/novel-coronavirus-2019/media-resources/press-briefings if you wish to be informed about what is effective.
It is interesting to see mathematics being used in this way to help with the analysis of Covid-19 – probably because I studied Pure Maths and Applied Maths myself in the past.
It all sounds a lot more solid and dependable than some of the Economics theories we discuss sometimes on here, Richard.
🙂
This technique has been used in Cambridge University to test every student weekly since the beginning of the University term. It works well, and reduces the total number of tests needed. More importantly it allows the test to be repeated at weekly intervals and that is the best way to push down the infection rate. Combined with a much better processing procedure that gives results very much more quickly it’s difficult to know how could be bettered
RWANDA develops a game changing testing algorithm, it is profoundly disappointing that PHE and their very expensive private sector consultants have ignored this development for so long.
There was a scientific paper published on the 30/04/2020 and n addition this was covered in July by BBC World Service – Science in Action.
https://www.bbc.co.uk/programmes/w3cszh0k
I’ve broken down the timings for the program below.
covid-19 mutations 1:37
Covid-19 testing algorithms 10:45
climate change 23:30
https://allafrica.com/stories/202005260054.html
By Ange Iliza
Rwandan medical researchers have discovered a new pooling algorithm of COVID-19 mass testing. It was conducted, approved and applied in Rwanda.
The newly discovered testing method combines samples, tests them in groups and bears the same results as other methods of mass testing.
The new method that has proven to be more effective was applied in a survey testing that led to easing lockdown on May 1st. It was conducted in 30 per cent of the health facilities across the 30 districts of Rwanda.
The research was published on Cornell University on April 30 after being worked from March and has proven to be effective.
Using the new method, the cost of mass testing could be reduced since one test kit can be used for several people. If infected individuals are quickly and effectively quarantined, the prevalence will fall and so will the costs of regularly testing everyone.
In the published research paper, the recent experiment screened 1,280 individuals using only 64 tests, an efficiency gain of 20.
The latter justifies the fall of COVID-19 curve since late April. Majority of new cases are imported and contained.
According to one of the researchers, Dr. Leon Mutesa, other countries such as South Africa are requesting that Rwanda assist them to apply the same algorithm.
Cost and time efficient
One of the factors that prompted the research was the cost and time it took to test. Both factors are essential in effectively fighting COVID-19.
Rapidly identifying and quarantining infected individuals is one of the most important strategies available to contain the virus. However, each diagnostic SARS-CoV-2 test costs 30-50 US dollars.
For the new algorithm, at low prevalence, this testing procedure uniquely identifies infected individuals in a small number of tests and therefore cuts the cost and time spent on average testing.
“We report proof of concept experiments in which a positive sample was detected even when diluted a hundred-fold with negative samples,” the published research paper reads in part.
The time factor of the new algorithm has proven to be even more suitable for a rapidly spreading infectious disease like COVID-19 because infected individuals who are still at large can infect others.
“Such a strategy provides a possible pathway to the long-term elimination of SARS-CoV-2.”
The first step in testing, swab collection, is labour intensive. The next step involves machines that require expensive chemical reagents, currently in short supply, as well as skilled personnel.
Rwanda has recorded a total of 327 COVID-19 cases since March 14 when the first case was reported. 237 of those have so far recovered which means there are only 90 active cases.
Thanks
So –It seems that to summarise then …
– Using maths is helpful (whether you aspire to it or not) — maths can also be a really useful tool to reveal patterns ( some mathematicians might find that offensive – I mean no harm).
– Using this maths insights can massively improve testing efficiencies (as long as the infection rate is low enough)
– This has been recognised in several places as an advantage (Rwandha, Israel, Cambridge University ….) – but not embraced more generally
– The UK claims to have world beating system that is ignoring a 90% cost reducation mechanism …. How must this make the people of Manchester feel?
– Let’s allow our world beating system explain their relative efficiencies — the standard has been defined for a few months…………. In testing and tracing. and the impact on reduced transmission….
Very informative and I enjoyed it, but pitched at a numerate audience. Thank you RM I will contact my MP and devolved health minister.
I was not aware of this application, but it is not surprising. For example, our computers would be incredibly slow if we did not adopt such efficiencies in designing software. To name just one – the Quicksort algorithm – makes it possible to order huge databases in a fraction of the time a one-at-a-time sort might need.
The bottle-neck, of a functioning Covid test system, has been the genetic test i.e. availability of lab reagents, instruments and trained technicians. It’s a disaster this approach has not been adopted, as it is so efficient when positivity rates are lower.
I can understand why private contractors might avoid alerting the client to this approach.
SAGE members, at least one within that large and august group, must have been aware of this. The Govt and its SAGE advisory group must be accountable and an inquiry determine if they have acted negligently.
One question arises, apart from the current administrative debacle. Can a single swab sample be divided into two or more samples without affecting the sensitivity, selectivity and reproducibility of the genetic id test (RT-PCR)? If so, there is no need to chase that individual for another swab one has another identical sample ready to re-test. That individual is only advised on their Covid state, they don’t need to know the mechanics.
The centralised model, for handing this pandemic, of Gove, Cummings and Johnson is disastrously falling apart. Fund local health resources and GPs to be the focus, to take back control of this new disease.
Uruguay and Rwanda are examples of less developed countries and how they showed rich economies how to manage the virus efficiently.
In the British Medical Journal “Uruguay is winning against covid-19. This is how … Pool testing” Sept 2020 https://www.bmj.com/content/370/bmj.m3575.short
The Rwandan strategy – “A strategy for finding people infected with SARS-CoV-2: optimizing pooled testing at low prevalence” https://arxiv.org/abs/2004.14934
Thanks
It begs the question as to why it is not being done widely here
I have been wondering for ages, and I mean months, why I haven’t heard the term “pool testing” in the news.
The testing of sewage which was discovered to be an accurate measure of infection in a community early on is a type of pool testing.
I have had this further idea which I had ages ago that may not work, but I think is kind of interesting. There are some good antigen tests coming online, but months ago there were antigen tests that were rejected because they were not accurate and gave false positives or negatives. What if you took a cheap mass produced antigen test before the PCR test and used it to increase the size of the PCR test pools? The antigen test does not have to be accurate, although accurate is better. It just has to have a known failure rate and be mostly accurate.
Instead of having random pools of a certain size, you have bigger pools of negative antigen tests. These would contain positives because the antigen test might be inaccurate but that doesn’t matter as long as the pool size can be increased.
Follow the money and we will probably find contractors are paid per test. Pooling samples reduces the pay.