3 ms·

Previous discussion and comment: https://news.ycombinator.com/item?id=28693855 https://news.ycombinator.com/item?id=28693855 I still think it's a bit tragic,

Previous discussion and comment: https://news.ycombinator.com/item?id=28693855 https://news.ycombinator.com/item?id=28693855 I still think it's a bit tragic, if they'd just abandoned the quantitative claims and stuck to +/- detection strategies (suitable for Covid testing, STD detection, etc) they might have become something that we'd be talking about positively today, a hero of the Covid outbreak because they were all set up to do comprehensive testing. That's what microfluidic approaches with small samples are good for, detecting presence/absence of a marker within a defined limit of detection. Anyone with even basic domain knowledge would have known that the quantitative stuff couldn't be done on microfluidic samples... It's a really strange story, as they got all these military-government people on board, and that led to more trust from outside investors, and it just kept rolling... yet anyone like myself or dozens of others with experience in that field could have told them it was all BS if consulted. Go figure.

carabiner 3y ago
I think the microfluidic testing could have worked with ML and computer vision. I believe there was/is an Israeli startup working on it a few years after Theranos.
- changoplatanero 3y ago
  I heard that the reason why it can't be done is that to get those quantitative measurements you have to react the blood sample with certain chemicals. To do multiple tests you need enough blood that you can divide it up into samples and react each one separately. So if you only start off with one drop its hopeless to run the full suite of tests.
- hn_throwaway_99 3y ago
  No, it is physically impossible to run quantitative tests on very small amounts of capillary blood. There have been studies (can dig them up but you can Google for them) that show that capillary blood simply is not homogenous enough at these one drop volumes. Meaning it doesn't matter how good your technology is, it's just not possible.
hn_throwaway_99 3y ago
There is nothing, zero, nada of value in Theranos' research. What difference would it have made if they just "stuck to +/- detection strategies"? All of these tests already exist, heck there are lots of at home tests you can already run with a few drops of blood from a finger prick (or, in the case of Covid, no blood at all).
- twelve40 3y ago
  so these delusions of grandeur is what's weird to me with these startups. I just got out of one of many years where the decision was made to double-down on the failed approach and "we'll find our customer eventually". That translates to many more years of headed in the wrong direction with nothing to show for it, all for a delusional hope to become a unicorn (no less) some day. Why can't people just settle for a safer bet after the failure becomes apparent? Perhaps if Theranos started selling more traditional tests, even like home tests, they would lose the unicorn status but they would still have a (smaller) business and people would not go to jail? But no, had to be a 100% scam and crash everything into the wall.
- photochemsyn 3y ago
  Well, perhaps not as far as known, but the general microfluidic platform they were working with was supposed to at least be capable of microfluidic capabilites (i.e. comparable to a DNA microfluidics system for DNA sequencing of the kind developed ~20 years ago). If that was working, now imagine a chip platform, a small wafer or plate containing ID sequences of the top 100 known human infectious diseases, in that situation possibly you could get a one-shot test using just 200 ul of blood, distributing 2 ul to each (non-quantitative, +/- test). Then you'd test it, see what your limit of detection is and so on. If this was automated and reliable, it'd be very valuable for many different purposes (you could also have specialized chips for all known respiratory viral types, all common water-borne pathogens, all known STDs, etc) Of course, that'd be a much smaller market, maybe only a fraction of the value of the whole quantitative blood component level testing industry.
  - hn_throwaway_99 3y ago
    Not sure if we are agreeing with each other or disagreeing, but what you are describing had absolutely nothing to do with what Theranos was actually researching and working on. So sure, I think it's possible to say "Man, if only Theranos was working on something totally different from what they were selling to investors and the public, there could have been something!"

carabiner 3y ago

I think the microfluidic testing could have worked with ML and computer vision. I believe there was/is an Israeli startup working on it a few years after Theranos.

changoplatanero 3y ago
I heard that the reason why it can't be done is that to get those quantitative measurements you have to react the blood sample with certain chemicals. To do multiple tests you need enough blood that you can divide it up into samples and react each one separately. So if you only start off with one drop its hopeless to run the full suite of tests.
hn_throwaway_99 3y ago
No, it is physically impossible to run quantitative tests on very small amounts of capillary blood. There have been studies (can dig them up but you can Google for them) that show that capillary blood simply is not homogenous enough at these one drop volumes. Meaning it doesn't matter how good your technology is, it's just not possible.

hn_throwaway_99 3y ago

There is nothing, zero, nada of value in Theranos' research. What difference would it have made if they just "stuck to +/- detection strategies"? All of these tests already exist, heck there are lots of at home tests you can already run with a few drops of blood from a finger prick (or, in the case of Covid, no blood at all).

twelve40 3y ago
so these delusions of grandeur is what's weird to me with these startups. I just got out of one of many years where the decision was made to double-down on the failed approach and "we'll find our customer eventually". That translates to many more years of headed in the wrong direction with nothing to show for it, all for a delusional hope to become a unicorn (no less) some day. Why can't people just settle for a safer bet after the failure becomes apparent? Perhaps if Theranos started selling more traditional tests, even like home tests, they would lose the unicorn status but they would still have a (smaller) business and people would not go to jail? But no, had to be a 100% scam and crash everything into the wall.
photochemsyn 3y ago
Well, perhaps not as far as known, but the general microfluidic platform they were working with was supposed to at least be capable of microfluidic capabilites (i.e. comparable to a DNA microfluidics system for DNA sequencing of the kind developed ~20 years ago). If that was working, now imagine a chip platform, a small wafer or plate containing ID sequences of the top 100 known human infectious diseases, in that situation possibly you could get a one-shot test using just 200 ul of blood, distributing 2 ul to each (non-quantitative, +/- test). Then you'd test it, see what your limit of detection is and so on. If this was automated and reliable, it'd be very valuable for many different purposes (you could also have specialized chips for all known respiratory viral types, all common water-borne pathogens, all known STDs, etc) Of course, that'd be a much smaller market, maybe only a fraction of the value of the whole quantitative blood component level testing industry.
- hn_throwaway_99 3y ago
  Not sure if we are agreeing with each other or disagreeing, but what you are describing had absolutely nothing to do with what Theranos was actually researching and working on. So sure, I think it's possible to say "Man, if only Theranos was working on something totally different from what they were selling to investors and the public, there could have been something!"