Medicine is plagued by untrustworthy clinical trials. How many studies are faked or flawed?

That's one of the things where AIs will be revolutionary. They will be able to analyze the entire evidence. All of it. Every paper and trial ever published, assess their reliability, find the duds, the frauds, the errors, the duplicates, and so on. In seconds. It will take more time for humans to review and digest the conclusions than it will take to do the full analysis.

Want to change some of the parameters of analysis, say how you "GRADE" the trials? Already done. Are you sure that open label trials with subjective outcomes are OK and their conclusions should be taken at face value? Because homeopathy looks convincing with those criteria, and so does a bunch of other nonsense. It won't be possible to get what is scientifically valid correctly and keep the nonsense in.

I predict that it will be absolutely brutal to "evidence-based medicine" in general, and even more so for the psychosomatic fringes that have ruined our lives. It will basically mark the entire field of biopsychosocial research as a bunch of crap that should never receive a single cent of funding again, maybe even call out our BPS overlords as basically frauds who keep doing the same stuff over and over again. Obviously with more diplomatic language.

I am very much looking forward to it. In general it will cause a lot of controversy with people unable to process that such a smart thing can be so wrong about something they are absolutely convinced of. Conspiracy theorists and people who espouse various ideologies in general. This includes psychosomatic ideologues, it will be very hard for many doctors to process this, this very same puzzlement as to why this smart thing is so wrong about something that is obviously true to them. It's not a coincidence that Internet trolls and conspiracy theorists LOVE everything about this "lol, it's just anxiety, bro"/"they're just faking it for the attention, damn malingerers get a job" stuff.
 
I think it is likely to be a lot worse than a quarter.

Though I doubt most of it is straight fraud. It is more like selective (and possibly sometimes unconscious) use of evidence and its interpretation and significance, massaging methodology to favour certain results or to limit or even eliminate the possibility of falsification from the start, etc. Enough to seriously bias the result, but not leave one too open to accusations of incompetence/misconduct/fraud.

A lot of it is going to be people just fooling themselves about their pet hypothesis, and being unable to admit it when it turns out they have fooled themselves. This particular problem will never go away, which is why robust methodology is so important.

See psychosomatics for clear examples of all the above.

rvallee said:
That's one of the things where AIs will be revolutionary.
Click to expand...
Just clarifying my comment to you the other day on AI. I wasn't suggesting that AI has no role, or only a minor one. Rather that we should not be putting too much faith in it, especially this early on in its development and application. It is just too open to GIGO for now.
 
You must log in or register to reply here.
Back
Top Bottom