efficient corona testing with limited resources

[Anonymous No.11528450]

So there are some ideas going around how to make testing for Corona more efficient, given resource and time limits on testing. One idea is to bulk test groups of people by simply mixing their bio samples, and quarantine everyone in case the test turns up positive. That way you'll be able to contain the virus with relatively few tests.

But this idea can be pushed further: what if we combine samples of lets say 100 people, and in case of positive, divide the group in two groups of 50, and re-test both groups, etc. Of course keep in mind that in this example, if a bulk test gives a positive, there can be more than one person who is positive.

So this can be solved analytically, but I suck at math: How do you find every infected person in a population of N people, while minimizing the number of tests? A safe assumption here is that if N is the total population of a country, there will be at least one person who's infected.

[Anonymous No.11528455]

fuck and piss and shit and cunt and fuck sucking nigger dicking cock CUNTING MOTHER FUCKING FUCK FUCK PISS FUCK SHIT CUNT FUCK FUCK FUCK

[Anonymous No.11528463]

>>11528455
not quite the response I was expecting but thanks for the bump anyway

[Anonymous No.11528536]

there's gotta be someone here who can solve this

[Anonymous No.11528580]

You obviously don't have an experience in biology. Get lost with these dumb threads.

[Anonymous No.11528588]

its coming

[Anonymous No.11528592]

Well, worst case is Nlog(N) tests with your method if everyone is infected. To get a realistic number, you need to have an estimate of the expected number of people infected.

[Anonymous No.11528663]

>>11528580
retard

>>11528592
Thanks! Predictions about infection numbers like this are possible to obtain with antibody measurements that are now underway. Basically the idea is to chart the proportion of the population that have been exposed to the virus, and track how it changes over time.

So maybe I should rephrase the question: what is the minimum number of tests you would need, given an estimated infection rate?

[Anonymous No.11528985]

I have no resources
I can drive 2km down there road and get a drive through test for free
sucks to be you op

[Anonymous No.11529100]

>>11528985
You misunderstand, I can get tested for free too. But resources for testing, like the primers for qPCR and sheer man hours that go into running a large number of tests etc, are simply limited. No matter where you are.

[Anonymous No.11529165]

>>11528985
>brags about having no resources
>not even a brain half as curious as cock sucking OPs
>tries to act superior in his fit of retardation
For what purpose do you continue to seek your next breath?

[Anonymous No.11529217]

>>11529100
What are you? Stupid? Do you think you can combine 100 probes and if one of them is infected, you'll get an amplification curve above the background noise? Have you ever ran a Rt-PCR test in your life?

[Anonymous No.11529286]

>>11529217
>hurr durr what is qPCR and what is upping the cycle numbers

[Anonymous No.11529381]

>>11529286
Yeah, let's base our decisions on an amplification curve that will get thrown out as a false positive in any other research. Thank god no one will ever put you in charge of anything.

[Anonymous No.11529387]

>>11529381
ok just shut the fuck up already

do I really need to spell it out for you why you are retarded?

[Anonymous No.11529418]

>>11529387
Spell out what, idiot? What does upping the cycles achieve? You still get an amplification curve that has a threshold barely above the background and the melting curve is all over the place. How much viral material do you think there's in one sample? Not to even start with the logistics of pipetting 100 samples into one PCR tube. How would you even calculate the master mix?

[Anonymous No.11529449]

>>11529418
In this hypothetical scenario we do not give a fuck about an accurate amplification curve, all we need is to determine is the presence of viral DNA.
>What does upping the cycles achieve?
What the fuck do you think? A threshold number of cycles is the clinical determinant for a positive sample and is used to infer viral load. Mix samples N times then true viral load is divided by N and can be compensated by setting clinical threshold to threshold + N.

>pipetting 100 sample
Jesus you really are retarded. You can soak the swabs simultaneously and pipet once.

[Anonymous No.11529451]

>>11529449
oh man it's getting professional in this bitch YES YES FUCK ME YES

[Anonymous No.11529522]

>>11529449
You seem to not get that the PCR hardware itself has limitations of confusing the detection of lower volumes of amplified DNA with the background noise.

Normally your finished PCR mix contains around 2.5 microliters of the sample. You want to do 100 samples in one tube? How are you gonna get 0,025 micro-litters out of a single sample?

I'm sure it's possible, but the places that have this sort of equipment and can execute the procedure, probably aren't complaining about mass testing in the first place.

[Anonymous No.11529569]

>>11529522
>PCR hardware itself has limitations of confusing the detection of lower volumes of amplified DNA with the background noise
Fucking hell. Singal intensity depends on the cycle number obviously, so will not be any different from a regular positive sample if the number of cycles is adjusted appropriately.

>You want to do 100 samples in one tube?
No, I want N samples in one tube. Obviously N cannot be unlimited. But it most definitely isn't limited to 1, as you seem to think.

Are we gonna keep jerking each other off or are you ready to admit that there are no unsolvable practical issues here?

[Anonymous No.11529609]

>>11529569

[Anonymous No.11529632]

>>11529569
>Singal intensity depends on the cycle number obviously

Yeah, in the false positive cycle range, especially if you want dilute even further the test sample. There's a reason you rarely see research do more than 40 cycles, even if that in theory would save a lot of the stock consumables. 10 samples could be doable, but the margin of error becomes too big with this 100 sample deal.

[Anonymous No.11529639]

>>11529632

[Anonymous No.11529678]

>>11528663
Forgot to reply before. I thought about this and can get exact numbers for N=2^m (i.e. a power of two) and p=2^-k, where N is the number of people, p is the infection rate, and k<=m. That is, we're assuming a random person has a chance of having the virus with probability p and that it's independent of anyone else being infected.
Lower bound: 2(N*2^-k + k) - 1
Upper bound: N(k+1)*2^(k-1) - 1

So we can see that, even in the worst case, this method is beneficial if k>=3 i.e. p<=0.125. At least with these assumptions.

You can do better than a binary search, though. I came across this: https://en.wikipedia.org/wiki/Group_testing#Generalised_binary-splitting_algorithm

[Anonymous No.11529689]

>>11529678
Oops, there's a typo in the upper bound result.
Upper bound: N(k+1)*2^-(k-1) - 1

You can get approximate answers for N not a power of two. Replace N with 2^log(N) should do it.

[Anonymous No.11529706]

>>11529632
>more than 40 cycles, even if that in theory would save a lot of the stock consumables

Actually, I'm an idiot. Wanting a longer amplification would actually require higher number of consumables. Still, the master mix is designed for the small size of a PCR tubes and around 40 cycles.

[Anonymous No.11529722]

>>11529678
>That is, we're assuming a random person has a chance of having the virus with probability p and that it's independent of anyone else being infected.
Oops, that's another mistake. I meant p is the fraction of infected people amoung the N people in the group. It's deterministic, not random.

[Anonymous No.11529727]

>>11529678
>>11529689
Thanks man! Won't have time for a detailed reply now but will do later. Anyway, I appreciate the reply

[Anonymous No.11529778]

>>11529706
>Actually, I'm an idiot.
No shit

[Anonymous No.11529792]

>>11529778
Nice argument. Attempt your little research and come back with the results so we can laugh you out of the board.

[Anonymous No.11529815]

>>11528450
>One idea is to bulk test groups of people by simply mixing their bio samples

That seems reasonable for households/families under the same roof. The rest of your proposition is just pointless though. Given the risk of false positives/negatives even with current testing kits and platforms having multiple tests per individual is the only way to really be sure who has it (and needs to be isolated + contact traced).

[Anonymous No.11529869]

>>11529792
>Nice argument
I was literally quoting your words buddy

[Anonymous No.11529919]

>>11529815
If the first part is reasonable then the rest should be too, it's a less extreme version of the second part.

[Anonymous No.11529992]

>>11529919
>then the rest should be too, it's a less extreme version of the second part.

It's not reasonable not only because of the pragmatic side but the practical limitations too (of which has already been discussed in this thread earlier). It just is not technologically feasible. What you are proposing is nothing more than a thought experiment at most. If I have on poisoned skittle in a bag of skittles and I have a test that can determine whether poison is present in a batch what is the fewest tests needed to find that one skittle in X numbers. That's all it is - as far as being implemented in a real world scenario is just plain stupid and full of flaws and pitfalls too great to get over with current technology. You'd be better off fobbing this question to a maths professor. I'm sure they;d find it amusing - I'm also sure this 'problem' or variation of it has been solved already as I'm pretty sure I've heard of it before.

[Anonymous No.11530072]

>>11529992
>If I have on poisoned skittle in a bag of skittles and I have a test that can determine whether poison is present in a batch what is the fewest tests needed to find that one skittle in X numbers. That's all it is
But it isn't.

[Anonymous No.11530076]

>>11529992
Lel, the arrogance of shitposters here is pretty amusing

[Anonymous No.11530465]

Bump

[Anonymous No.11531578]

Stayin alive