They failed on several fronts

1. The basic equipment they built simply didn't work at all

2. Their workaround trick of getting a finger prick blood draw, dilute it with water and run the test on a Qwest machine produced wrong results.

3. The fundamental approach itself was unviable. You can't get reliable results from a finger prick.

Theranos failed on all fronts except being able to attract investments from non experts.

So, the first issue (detection) comes down to the detection of a target gene at some minimal level using polymerase chain reaction or perhaps an enzyme-binding assay or something like that. These molecular biology techniques can actually detect even a few copies of a gene (i.e. a virus) in a sample tube, so a finger prick could plausibly work for this kind of detection goal. The main thing to keep in mind is the difficultly of quantitative measurements at these low levels. For example, I wouldn't be easily able to tell if my sample had ten copies of the gene or a hundred copies of the gene, certainly not cheaply and efficiently.

But, the blood chemistry tests - the second, quantitative goal - are far worse. Those are supposed to produce results with error bars, I'd guess at least +/- 10% accuracy is what you want (I'm not a doctor or nurse). But if you have a finger prick, well, what about evaporation of liquid volume? You can see it happening on a really good scale, 10 ul will evaporate away pretty fast. Also, is a finger prick (capillary) going to be like blood from a vein?

The problem is that precision with tiny volumes is possible, but requires massive preparation and dedicated professionals and certainly doesn't scale easily. Any such system would have to be calibrated daily, tested (with known standards), then samples would be run, then you test it again... Let alone cross contamination and about a hundred other problems. Furthermore - 26 different quantitative tests at once???

So, that's the kind of thing I meant when talking about the easier detection goal (where 26 pos/neg 'is the gene there' tests on a single sample might be plausible) vs. the much harder quantitative goal.

There is room for improvement in testing technology and processes.

But not to the extent of "One drop of blood from your finger can provide accurate tests on 26 different types of tests" Especially when some of those tests are only accurate with a large sample of blood

I'm always reminded of this when Theranos comes up. Years ago I was in the hospital, they wouldn't let me out until a blood test showed some numbers that made them feel better. I asked about how the one test worked. The nurse would come in and draw a tube of blood (I forget the volume, but if you've had your blood drawn for a test, a typical tube sized - 10's of mL). And then that'd go down to the lab and be mixed with reagents and a tech would look in a microscope slide and count things. Then they'd reverse the math and turn the things they counted into a concentration of stuff in my blood. That's the number the doctor looked at. So when I worked through the process, and deduced that what the lab tech counted was on average 4 particles of interest. And since I was an obnoxious grad student, I brought up shot noise and said that means there's a 50% error in the measurement and that the doctor should discharge me. Of course, he said that even with a 50% buffer, the numbers were too low for him to feel comfortable.

Long story, but they needed an entire tube of my blood to count 4 particles in the microscope. A drop of blood would not have enough things in it to count. Real life is discrete, and often we need a big sample just to get enough things to count.

That's why I always though Theranos was a fraud.

Were they the first ones to do it?