3 ms·
The full process as described in the book is even more shocking. Theranos would dilute the finger stick draws, and then run them on the Siemens machines 5 time
by cepth 8y ago
The full process as described in the book is even more shocking.
Theranos would dilute the finger stick draws, and then run them on the Siemens machines 5 times. In theory, you could then calculate the concentration, and adjust for the dilution. In reality, the diluted samples would have concentrations of the relevant cells and elements that were too low to pick up.
To “fix” this, they would do those 5 runs, then calculate the median. And, if you had 5 outliers in all different directions, they’d discard the run and do 5 again.
The fact that PhDs and doctors who swore the Hippocratic oath allowed this to go on for over a year is shocking.
- mikekchar 8y agoI don't know much about Theranos, but I remember writing a proof in university. The TA in the class was a PhD student doing research in group theory. My proof relied on doing some manipulation of exponents and there was a point where I was doing "1 + (-1) = 0". The TA just could not understand that bit. He was absolutely sure that in some circumstances this wouldn't hold. No amount of discussion about how integers form a field under addition and multiplication would convince him. He eventually got his PhD and I can only assume he's working for a company like Theranos. Cue someone replying and telling me why I was actually the one that was wrong ;-) But the point is that even well educated people make really, really stupid mistakes.
- cepth 8y agoFair enough! The book is worth the read. I think I may be being a bit unfair to the scientists. Basically, Theranos was a hot enough startup that Elizabeth Holmes and Sunny Balwani would fire, demote, or marginalize anyone who made a fuss. Lifelong academics may have never had to deal with that kind of bullying/pressure before. On the other hand, they had some pretty incredible scientific talent on staff. You would think that people who are highly cited and have a reputation to protect would have spoke up sooner. Instead, the first two people to make any kind of sustained fuss were two fresh out of college hires.
- busterarm 8y agoAre we sure about this? My impression of this lately is that to be a lifelong academic is to constantly put up with bullying and pressure. Such is the nature of competing in a status-economy.
- Fomite 8y agoThere are definitely some labs I know people who have worked in that would give the most abusive startup a run for their money.
- flukus 8y ago> On the other hand, they had some pretty incredible scientific talent on staff. You would think that people who are highly cited and have a reputation to protect would have spoke up sooner. Instead, the first two people to make any kind of sustained fuss were two fresh out of college hires. Knowing how to work within the system is probably how they were afforded the chance to build this reputation int the first place. It's possible they were trying to do the same here but didn't realize the fundamental differences between the academic system and the commercial one. Fresh college grads haven't yet learned how to "not make waves", which is a blessing and a curse.
- whatshisface 8y agoThe road to correct intuition starts at wrong intuition and passes through no intuition - that's probably where the TA was in his professional journey. As for the statistical test, it goes beyond "mistake" and steps in to being something so unorthodox that any scientist who did it would have to be consciously wiggling. Even a group of children will complain if you're playing a board game and try to re-roll the dice unil you get to move forwards the number of squares that your investors would like.
- Bartweiss 8y agoAlso alarming - the Siemens machines already dilute samples, and the amount of dilution is set (in part) by making sure every sample test will have a usable concentration. So Theranos didn't just dilute where others didn't, they took a system that already used up all of the 'safe' margin for dilution and knowingly added more!