Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AI cannot be used to deny health care coverage, feds clarify to insurers

arstechnica.com

Post reply on HN
Print view View original

51–55 of 55 posts

Re: AI cannot be used to deny health care coverage, feds clarify to insurers

#51

Post by mindslight »

singleshot_ wrote: post #28
mindslight wrote: post #27

Earlier quoted context omitted.

Placing an IV after they had already botched it three times, living on lactated Ringer's instead of real food for the better part of a day, IV sedation anesthesia and its possible complications, (I didn't investigate side effects of the contrast), and the further destruction of bodily autonomy and resulting lack of alertness that comes from all those.

Yikes. Seems like a pretty serious MRI. Good luck. On one hand, this defensive practice of medicine is objectionable. On the other, it only exists because people like to sue providers when outcomes are bad. I’m glad you had a good outcome.

Everything turned out fine, thank you. I do get why in an emergency context doctors want to do as many tests as possible in parallel, and have very little context to guide them otherwise. But also I don't have much sympathy for things "only" being a certain way because of some other thing. The medical industry is a whole set of self-reinforcing terrible dynamics, which is why we need sweeping reform.

I've also got to wonder how much this narrative bemoaning MRIs for "headaches" is because doctors have rightfully moved to doing less harmful MRIs instead of less expensive X-ray CTs.

Re: AI cannot be used to deny health care coverage, feds clarify to insurers

#53

Post by faeriechangling »

sirspacey wrote: post #7
pylua wrote: post #4

I don’t see how we can use ai and not have it generate new biases. I also feel like biases have been baked into insurance data in a discriminatory way for a long time — think about how much car insurance is for a 16 year old boy.

Yes but consider the alternative - how will we eliminate bias without it? Having worked in medical billing, it is very scary how medicine as an applied science is carrying around a huge ball and chain of “insurance hasn’t figured out that their medical model is decades out of date.” Big, complex, data intensify problems are not going to be solved by an army of humans on legacy ERPs.

I cannot even describe how many problems are caused by medicine being shackled to the whims of insurers in terms of how stultifying it is and how many bad incentives it creates to workaround the limitations of insurance.

No idea how to fix the problem though.

Re: AI cannot be used to deny health care coverage, feds clarify to insurers

#54

Post by jwilber »

DistractionRect wrote: post #45
anon373839 wrote: post #40

Earlier quoted context omitted.

I believe it’s possible with decision trees and forests, even though they’re very complex inside, to calculate the contribution that each independent variable made to the final result. Of course, I suppose those haven’t been deemed “AI” for a few years.

Sure, you can step through it and see what's occurring, but there's zero explanation to accompany it. To fully reverse engineer and explain a reasonably complex AI model would be quite the undertaking. Instead, we have prompt engineering. It's easier to figure out how to ask the black box nicely/properly to do what you want than to open it up and optimize/tune it proper. If we assume that we get that far, I fully exp…

The explanation is unbelievably straightforward for tree models, even if you don’t fully understand entropy/information gain? What?

Re: AI cannot be used to deny health care coverage, feds clarify to insurers

#55

Post by sirspacey »

faeriechangling wrote: post #53
sirspacey wrote: post #7

Earlier quoted context omitted.

Yes but consider the alternative - how will we eliminate bias without it? Having worked in medical billing, it is very scary how medicine as an applied science is carrying around a huge ball and chain of “insurance hasn’t figured out that their medical model is decades out of date.” Big, complex, data intensify problems are not going to be solved by an army of humans on legacy ERPs.

I cannot even describe how many problems are caused by medicine being shackled to the whims of insurers in terms of how stultifying it is and how many bad incentives it creates to workaround the limitations of insurance. No idea how to fix the problem though.

Like every single public good I’ve investigate for social entrepreneurship the answer is:

Start an insurance company

It sounds impossible, but every one of these giants were founded in the past century.

They can be beaten with enough capital, power and ingenuity.

But that’s not gonna be done by a pre-seed startup with $1M in the bank and some cool products.

Genuine change requires a long view and a willingness to compete against a cartel.

Post reply on HN