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

ChatGPT Health

openai.com

Post reply on HN
Print view View original

461–470 of 677 posts

Re: ChatGPT Health

#461

Post by diydsp »

Rebuff5007 wrote: post #379
mlmonkey wrote: post #156

My uncle had an issue with his balance and slurred speech. Doctors claimed dementia and sent him home. It kept becoming worse and worse. Then one day I entered the symptoms in ChatGPT (or was it Gemini?) and asked it for the top 3 hypotheses. The first one was related to dementia. The second was something else (I forget the long name). I took all 3 to his primary care doc who had kept ignoring the problem, and asked…

I've heard a lot of such anecdotes. I'm not saying its ill-intentioned, but the skeptic in me is cautious that this is the type of reasoning which propels the anti-vax movement. I wish / hope the medical community will address stories like this before people lose trust in them entirely. How frequent are mis-diagnosis like this? How often is "user research" helping or hurting the process of getting good health outcome…

I can see why, but this is doc+patient in collab. And driven by using science in the form of applying llm-as-database-of-symptoms-and-treatments.

Anti-vax otoh is driven by ignorance and failure to trust science in the form of neither doctors, nor new types of science. Plus, anti-vax works like flat earth; a signaling mechanism of poor epostemic judgment."

Re: ChatGPT Health

#462

Post by lionkor »

All the Americans here arguing why this is a good thing, how your system is so flawed, etc. remember that this will be accessible to people in countries with good, free healthcare.

This is going to be the alternative to going to a doctor that is 10 minutes by car away, that is entirely and completely free, and who knows me, my history, and has a couple degrees. People are going to choose asking ChatGPT instead of their local doctor who is not only cheaper(!!!) but also actually educated.

People saying that this is good because the US system specifically is so messed up and useless are missing that the US makes up ~5% of the world's population, yet you think that a medical tool made for the issues of 5% of the population will be AMAZING and LIFE SAVING for the other 95%, more than harmful? Get a grip.

Not to mention shitty doctors, which exist everywhere, likely using this instead of their own brains. Great work guys.

I suspect the rationale at OpenAI at the moment is "If we don't do it, someone else will!", which I last heard in an interview with someone who produces and sells fentanyl.

Re: ChatGPT Health

#463

Post by ogogmad »

sensanaty wrote: post #305
cyberpunk wrote: post #91

Earlier quoted context omitted.

Yep same, with the caveat that any actionable advice requires actual research from reliable sources afterwards (or at least making it cite sources). i mean; i kinda get the concerns about misleading people but … are people really that dumb? okay if it’s telling you to drink more water, common sense. If you’re scrubbing up to perform an at home leg amputation because it misidentified a bruise then that’s really on you…

> are people really that dumb? Yes, absolutely. The US has measles back in rotation because people are "self-educating" (aka taking to heart whatever idiocy they read online without a 2nd thought), and you think people self diagnosing with a sycophant sentence generator is anything but a recipe for disaster?

ChatGPT frequently pushes back against the things I say. I think you can dial down the level of sycophancy.

Re: ChatGPT Health

#464

Post by ogogmad »

lm28469 wrote: post #331
drusepth wrote: post #177

Earlier quoted context omitted.

This isn't feasible for a huge swathe of the USA, often because of costs/insurance but sometimes literally just accessibility/availability. A few years ago it took me nearly 8 months to find a PCP in my city that was accepting new patients (and, wee, they dropped my insurance less than a year after).

> often because of costs/insurance but sometimes literally just accessibility/availability. These are self inflicted problems, we should work on these and improve them, not give up and rely on llms for everything

Is there a proven and guaranteed way to do this? Because otherwise it sounds very idealistic, almost like "if everything were somehow better, then things would be less bad". Doctor time will always be scarce. It sounds like it delays helping people in the here and now in order to solve some very complicated system-wide problem.

LLMs might make doctors cheaper (and reduce their pay) by lowering demand for them. The law of supply and demand then implies that care will be cheaper. Do we not want cheaper care? Similarly, LLMs reduce the backlog, so patients who do need to see a doctor can be seen faster, and they don't need as many visits.

LLMs can also break the stranglehold of medical schools: It's easier to become an auto-didact using an LLM since an LLM can act like a personal tutor, by answering questions about the medical field directly.

LLMs might be one of the most important technologies in medicine.

Re: ChatGPT Health

#465

Post by smeej »

cael450 wrote: post #437
ToucanLoucan wrote: post #406

Earlier quoted context omitted.

> How frequent are mis-diagnosis like this? The anecdote in question is not about mis-diagnosis, it's about a delayed diagnosis. And yeah, the inquiry sent a doctor down three paths, one of which led to a diagnosis, so let's be clear: no, the doctor didn't get it completely on their own, and: ChatGPT was, at best, 33% correct. The biggest problem in medicine right now (that's creating a lot of the issues people have…

I'm married to a provider. It is absolutely insane what she has to do for insurance. She's not a doctor, but she oversees extensive therapy for 5-10 kids at a time. Insurance companies completely dictate what she can and can't do, and frequently she is unable to do more in-depth, best-practice analysis because insurance won't pay for it. So her industry ends up doing a lot of therapy based on educated guesswork. Ever…

> Insurance companies completely dictate what she can and can't do, and frequently she is unable to do more in-depth, best-practice analysis because insurance won't pay for it.

The distinction between "can't do" and "can't get paid for" seems to get lost a lot with medical providers. I'm not saying this is necessarily what's happening with your wife, but I've had it happen to me where someone says, "I can't do this test. Your insurance won't pay for it," and then I ask what it costs and it's a few hundred or a couple thousand dollars and I say, "That's OK. I'll just pay for the test myself," and something short-circuits and they still can't understand that they can do it.

The most egregious example was a prescription I needed that my insurance wouldn't approve. It was $49 without insurance. But the pharmacy wouldn't sell it to me even though my doctor had prescribed it because they couldn't figure out how to take my money directly when I did have insurance.

I get that when insurance doesn't cover something, most patients won't opt to pay for it anyway, but it feels like we need more reminders on both the patient and the provider side that this doesn't mean it can't be done.

Re: ChatGPT Health

#466

Post by throwmeaway820 »

exabrial wrote: post #440

Going to a probabilistic system for something that can/should be deterministic sets off a lot of red flags. I’ve worked medical software packages, specifically a drug interaction checker for hospitals. The system cannot be written like a social media website… it has to fail by default, and only succeed when an exact correct solution was determined. This result must be repeatable given the same inputs. The consequence…

a drug interaction checker can be deterministic, based on a static corpus of drug interaction data

a diagnostic system should not necessarily be deterministic, because it always operates on incomplete data and it necessarily produces estimates of probability as an output

Re: ChatGPT Health

#467

Post by destitude »

Rebuff5007 wrote: post #379
mlmonkey wrote: post #156

My uncle had an issue with his balance and slurred speech. Doctors claimed dementia and sent him home. It kept becoming worse and worse. Then one day I entered the symptoms in ChatGPT (or was it Gemini?) and asked it for the top 3 hypotheses. The first one was related to dementia. The second was something else (I forget the long name). I took all 3 to his primary care doc who had kept ignoring the problem, and asked…

I've heard a lot of such anecdotes. I'm not saying its ill-intentioned, but the skeptic in me is cautious that this is the type of reasoning which propels the anti-vax movement. I wish / hope the medical community will address stories like this before people lose trust in them entirely. How frequent are mis-diagnosis like this? How often is "user research" helping or hurting the process of getting good health outcome…

You must not be involved in the medical field to realize how bad it is especially when it come to diagnosis.

Re: ChatGPT Health

#468

Post by NuclearPM »

Vegenoid wrote: post #308
NuclearPM wrote: post #148

Earlier quoted context omitted.

Brusqueness? More like insensitivity, lack of empathy, and ignorance. My 12 year old daughter (correctly) diagnosed herself to a food allergy after multiple trips to the ER for stomach pains that resulted in “a few Tylenol/Advil with a glass of water”.

This isn't a criticism of you, I don't know your full story. But I think many people have a misconception of the role of an ER. I know an ER doctor well, and the role of an ER is to, in approximate order of priority: 1. Prevent someone from dying 2. Treat severe injuries 3. Identify if what someone is experiencing is life-threatening or requires immediate treatment to prevent their condition worsening 4. Provide basi…

Where I live there is a lack of family doctors.

Re: ChatGPT Health

#469

Post by TZubiri »

threethirtytwo wrote: post #429
websiteapi wrote: post #423

Earlier quoted context omitted.

> The proper decorum here is if the doctor made the wrong diagnosis. All fees and causal charges made by the doctor must be fully refunded and paid for. It’s only fair given the premium they were originally given to make a false diagnosis. lol terrible idea. just as great as having so that the service you bought is entirely refunded if the code has a single bug.

Well the issue here is that the bug can cause you to die or can fuck up your entire life. A software bug generally doesn’t do anything to you and you actually don’t pay much money for software. One false diagnosis from a doctor costs you thousands of dollars and fucks up your life. Remember mcas? The bug on the 737 max that forced Boeing to pay reparations? That’s the level of bullshit people are dealing with for doc…

Legally speaking, why should the rules for important services be different than unimportant services. The price paid for the services is proportional to its value, if there are no consequences for a bug in your code, it was worth less and you should have paid less for it, the refund would be commesurate.

It's worth noting that you framed the discussion in terms of refunds, so any extra human life uncalculable value isn't really within the scope of a refund, you'd have a malpractice case which is entirely different from a breach of contract. This is just about the fees paid for the service.

Re: ChatGPT Health

#470

Post by steveridout »

exabrial wrote: post #440

Going to a probabilistic system for something that can/should be deterministic sets off a lot of red flags. I’ve worked medical software packages, specifically a drug interaction checker for hospitals. The system cannot be written like a social media website… it has to fail by default, and only succeed when an exact correct solution was determined. This result must be repeatable given the same inputs. The consequence…

Human doctors are probabilistic systems.
Post reply on HN