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

31–40 of 55 posts

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

#31

Post by DistractionRect »

vermarish wrote: post #29
vitehozonage wrote: post #11

Earlier quoted context omitted.

All regulations involving AI just make me wonder how the hell anyone can draw a line between statistics and whatever ""AI"" means in 2024. Getting angry at the moment every time i see "AI" because the terminology is just nonsense to me. Is it just neural networks now?

"Statistics" = models with interpretable decision outcomes "AI" = models without It's just a little bit easier to hold someone accountable for a logistic regression with predictive bias than for a neural network with similar issues.

Exactly this.

Decisions based on statistical models can be fully audited and understood, and the results are reproducible because the algorithm is well documented.

The same can't be said of AI models.

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

#32

Post by drc500free »

RCitronsBroker wrote: post #6
gumby wrote: post #3

Instead, a physician working for the insurance company will use some sort of GPT to come up with an excuse to deny care and then they will write that up in their own words to deny care.

as a med student, those guys are an embarrassment to us all. I had to deal with those guys to get my insurance to cover recombinant antibodies for the treatment of several autoimmune diseases. They haven’t practiced clinically for a decade, haven’t kept up with research ever since they left residency, but still don’t see any problem with denying anything but the cheapest immunosuppressant, no matter how bad the side…

Medical ethics really should preclude what these insurance company physicians are doing, which is sometimes rubber-stamping 5000 denials a day. Maybe an organized campaign of board complaints is in order.

Propublica has done an excellent slate of investigatory articles recently about how insurance denials actually work.

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

#33

Post by NPC82 »

earthwalker99 wrote: post #22
RCitronsBroker wrote: post #6

Earlier quoted context omitted.

as a med student, those guys are an embarrassment to us all. I had to deal with those guys to get my insurance to cover recombinant antibodies for the treatment of several autoimmune diseases. They haven’t practiced clinically for a decade, haven’t kept up with research ever since they left residency, but still don’t see any problem with denying anything but the cheapest immunosuppressant, no matter how bad the side…

[flagged]

The physician who gives legitimacy to denying coverage at an insurance company is at the very heart of what you're complaining about. As the other poster has pointed out, and many physicians online, they are so out of practice and out of touch they basically say whatever the insurance company wants them to for money.

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

#34

Post by ProllyInfamous »

RCitronsBroker wrote: post #6
gumby wrote: post #3

Instead, a physician working for the insurance company will use some sort of GPT to come up with an excuse to deny care and then they will write that up in their own words to deny care.

as a med student, those guys are an embarrassment to us all. I had to deal with those guys to get my insurance to cover recombinant antibodies for the treatment of several autoimmune diseases. They haven’t practiced clinically for a decade, haven’t kept up with research ever since they left residency, but still don’t see any problem with denying anything but the cheapest immunosuppressant, no matter how bad the side…

As a med student DROP OUT, I want to encourage you to keep this soul as you slowly enter practice.

Oftentimes, the "embarrassment to us all" is just trying to pay his own bills/obligations [not saying this is "correct" — it's not].

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

#36

Post by bonton89 »

g42gregory wrote: post #35

I think all it means that you need a person to push the final button. Am I wrong?

Whew, I was afraid this AI thing was going to mean we were all out of a job. But machines cannot replace the human ability to take the blame. You need a warm body for that.

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

#37

Post by bonton89 »

nullindividual wrote: post #14
SOLAR_FIELDS wrote: post #12

Earlier quoted context omitted.

I think that’s ok from the perspective of the discussion at hand because we are still talking about an actual human’s terrible decision having their name attached to it. I think the system used to deny care being comically broken is a wholly different issue.

Insurers now refuse to release the name of the contracted doctor who made the denial.

That's one way to prevent a malpractice suit I guess.

Why doesn't the AMA expel these doctors from its ranks?

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

#38

Post by tzs »

vermarish wrote: post #29
vitehozonage wrote: post #11

Earlier quoted context omitted.

All regulations involving AI just make me wonder how the hell anyone can draw a line between statistics and whatever ""AI"" means in 2024. Getting angry at the moment every time i see "AI" because the terminology is just nonsense to me. Is it just neural networks now?

"Statistics" = models with interpretable decision outcomes "AI" = models without It's just a little bit easier to hold someone accountable for a logistic regression with predictive bias than for a neural network with similar issues.

As long as the model is deterministic in operation, even if the process of building the model is non-deterministic, can't decisions be interpreted by rerunning the model with tweaked inputs to find what would be needed to change the outcome?

For example suppose a bank rejects a credit application and they need to tell the applicant why. Repeatedly add $1000 to the income input and check again until until the model approves, and then tell the application they were rejected for insufficient income and how much more they would need to pass.

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

#39

Post by vermarish »

tzs wrote: post #38
vermarish wrote: post #29

Earlier quoted context omitted.

"Statistics" = models with interpretable decision outcomes "AI" = models without It's just a little bit easier to hold someone accountable for a logistic regression with predictive bias than for a neural network with similar issues.

As long as the model is deterministic in operation, even if the process of building the model is non-deterministic, can't decisions be interpreted by rerunning the model with tweaked inputs to find what would be needed to change the outcome? For example suppose a bank rejects a credit application and they need to tell the applicant why. Repeatedly add $1000 to the income input and check again until until the model ap…

Yes, this is a counterfactual line of reasoning. It's certainly legit, and I believe it's an active field of research in ML, but at the end of the day, you're still poking around the inputs of a black-box model as opposed to examining a white-box model.

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

#40

Post by anon373839 »

DistractionRect wrote: post #31
vermarish wrote: post #29

Earlier quoted context omitted.

"Statistics" = models with interpretable decision outcomes "AI" = models without It's just a little bit easier to hold someone accountable for a logistic regression with predictive bias than for a neural network with similar issues.

Exactly this. Decisions based on statistical models can be fully audited and understood, and the results are reproducible because the algorithm is well documented. The same can't be said of AI models.

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.
Post reply on HN