3 ms·

Also, their genetic risk scores and population admixtures are really bad. I can't understand why they are so bad given that they hired pretty good researchers a

by nextos 2y ago

Also, their genetic risk scores and population admixtures are really bad. I can't understand why they are so bad given that they hired pretty good researchers and building these is quite simple. Freely available models run circles around anything they report on their site. It's a bit like uBiome, they have sold a lot of snake oil and harmed the reputation of B2C tests. It's a shame as something like 23andme, plus a bit of epigenetic testing to capture environmental factors, could be a wonderful way to get an overall health snapshot.

Glyptodon 2y ago
Well... My experience (having worked as an SWE) w/ medical technology is that if a company is selling something, they will choose the version of advice or analysis that most aligns with selling something. (I got ordered to adjust scoring thresholds in a statistical grouping to have "nicer" groups, for example.) If your company does treatment X that competes with treatment Y, they'll look to expenand the edge cases that suggest X over Y as much as possible. If a company wants people to feel like they're getting something out of a genetic profile, they'll report the broadest version of risk, and then slow roll more detailed analysis. Things like this are why I strongly think certain profit motives and business models should be extremely restricted. Just like private prisons create a profit motive for creating crimes, medical services have a profit motive for spreading inaccurate and twisted medical advice, whether it's things like alternative or new age medicine, treatment modality choices, or DNA information.
- WalterBright 2y ago
  Government funded research also has perverse incentives: 1. publish or perish, leading to lots of low quality papers 2. funding doesn't continue if one doesn't get results, leading to selection of "safe" research rather than risky research, and results that cannot be replicated 3. no funding for politically unpopular topics and, of course, the reasons why people publish overtly fraudulent research papers.

by nextos 2y ago

Glyptodon 2y ago

Well... My experience (having worked as an SWE) w/ medical technology is that if a company is selling something, they will choose the version of advice or analysis that most aligns with selling something. (I got ordered to adjust scoring thresholds in a statistical grouping to have "nicer" groups, for example.) If your company does treatment X that competes with treatment Y, they'll look to expenand the edge cases that suggest X over Y as much as possible. If a company wants people to feel like they're getting something out of a genetic profile, they'll report the broadest version of risk, and then slow roll more detailed analysis. Things like this are why I strongly think certain profit motives and business models should be extremely restricted. Just like private prisons create a profit motive for creating crimes, medical services have a profit motive for spreading inaccurate and twisted medical advice, whether it's things like alternative or new age medicine, treatment modality choices, or DNA information.

WalterBright 2y ago
Government funded research also has perverse incentives: 1. publish or perish, leading to lots of low quality papers 2. funding doesn't continue if one doesn't get results, leading to selection of "safe" research rather than risky research, and results that cannot be replicated 3. no funding for politically unpopular topics and, of course, the reasons why people publish overtly fraudulent research papers.