Live data from Hacker News

I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

inflecthealth.medium.com

221–230 of 368 posts

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#221
post #189

Earlier quoted context omitted.

A half-working system is cute and useful when it recommends a pizza place you might like. But the same system is a menace to society when you tell it to drive a car or prescribe a drug to treat an illness or decide a person's prison sentence or protect a perimeter with lethal force. It's absolutely amazing that so many AI scientists don't seem to understand this distinction.

What about helping correct the grammar and style of an article? Writing out boring code? Creating a test for code you already have? These are all tasks that have important economic consequences. AI is not half-bad at them, it is actually approaching human levels of performance. If we continue to improve AI, maybe one day its performance in these critical tasks you describe will also surpass humans.

Is it actually approaching human levels of performance, or are we continually reducing the commercially acceptable skill level of all of these things?

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#222
The prompt matters, and GPT-4 is better than ChatGPT (apologies for missing line breaks, but the point is still clear):

Prompt: A 26 year old female patient presents at the emergency room with lower right quadrant pain. What are the question you would ask her and what is the differential diagnosis?

Response: As an AI language model, I'm not a doctor, but I can suggest some questions you could ask the patient and offer possible differential diagnoses. It is important that you consult a healthcare professional for proper assessment and diagnosis.

Questions to ask the patient:

When did the pain start? Can you describe the nature of the pain (e.g., sharp, dull, cramping)? Has the pain changed in intensity or location since it began? Have you experienced any fever, nausea, or vomiting? Are you experiencing any changes in bowel habits, such as diarrhea or constipation? Have you had any urinary symptoms, such as pain during urination or increased frequency? Are you currently menstruating or have you recently missed a period? Is there any possibility that you may be pregnant? Have you experienced this type of pain before? Are there any other associated symptoms, such as dizziness or shortness of breath? Do you have any significant past medical or surgical history? Are you currently taking any medications, including over-the-counter or herbal remedies? Possible differential diagnoses for lower right quadrant pain in a 26-year-old female include:

Appendicitis: Inflammation of the appendix, typically characterized by pain that starts around the belly button and migrates to the lower right abdomen. Ovarian cyst rupture or torsion: A cyst on the ovary can cause pain if it ruptures or twists. Ectopic pregnancy: A pregnancy that implants outside the uterus, usually in the fallopian tube, can cause severe abdominal pain and should be ruled out in women of childbearing age. Pelvic inflammatory disease (PID): Infection of the female reproductive organs, often caused by sexually transmitted infections, can lead to lower abdominal pain. Kidney stones: Pain from a kidney stone can radiate to the lower abdomen or groin. Gastroenteritis: Inflammation of the stomach and intestines, often from a viral or bacterial infection, can cause abdominal pain and cramping. Inflammatory bowel disease (Crohn's disease or ulcerative colitis): Chronic inflammation of the gastrointestinal tract can cause abdominal pain, diarrhea, and other symptoms. Irritable bowel syndrome (IBS): A functional gastrointestinal disorder that can cause abdominal pain, bloating, and changes in bowel habits. Urinary tract infection (UTI): An infection in the urinary system can cause pain, often accompanied by increased urinary frequency or pain during urination.

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#223
post #206

Earlier quoted context omitted.

I my mind, the AI only needs to be more reliable than the human expert, and doctors and lawyers and other experts are "only human". There is only so much a human can read and remember. There is only so much they can be convinced to change their mind in the face of new evidence. They get tired. They get cranky. They have biases, and fears. Some doctors are just terrible. I think the AI is probably more reliable than t…

I suspect an ER doctor who misdiagnosed patients 50% of the time would not be an ER doctor for very long. Also consider that, in this case, the 50% the AI got right was defined such somewhat generously. For each patient, the AI presented several possible diagnoses, only one of which was correct. If the AI were making all the decisions, would the AI also decide to treat the correct ailment in all those cases? In an em…

But it's also a ridiculous assessment. How much training material for ER doctors was in the training set? I wonder what my rate would be (having never done a single course in medicine)... It should be totally unsurprising that it produced bad results. That it got as much right is an indication there's a good chance you can do reasonably well by training a model with a lot of actual case studies.

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#224
> My fear is that countless people are already using ChatGPT to medically diagnose themselves rather than see a physician.

My fear is that professionals will start to use ChatGPT too liberally to augment or multiply their work in cases like this. The danger here might be like the autopilot problem... i.e The idea of staying alert and focused on the road while counter-intuitively not participating is nearly humanly impossible. If ChatGPT is used as the autopilot of certain professions, things will begin to be missed, even though we know it's highly fallible - it's difficult to vet every single response in detail with a critical eye.

One reasonable argument is that for areas severely lacking in human workers the average might be a net positive, but the overall quality will be reduced.

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#225
I found this quote really interesting:

> this is not an uncommon scenario; about 8% of pregnancies discovered in the ER are of women who report that they’re not sexually active.

We have so much work to do as a society to get honest about our bodies. Hoping my children do better; they are already getting better education than my wife did.

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#226

Earlier quoted context omitted.

> ChatGPT does not ask for clarification, because it is not a knowledge base trying to find an answer. All it does is figure out what character is statistically most likely to come next, it has no heuristic to know that there is a task it hasn't fully completed. It absolutely can, but it requires the right prompts. If you just say "give me a diagnosis" it won't, but if you prompt it with a "you are a doctor and I am…

ChatGPT's heuristic may lead it to emit a sequence of words that suggest to English speakers that they should provide clarification, but ChatGPT is congenitally incapable of determining that answering a question requires more information that was provided in the prompt and that clarification is needed. Rephrased, ChatGPT can ask for clarification, but only as a side effect of satisfying it's heuristic.

You can ask it to complain if there is a lack of information in the query. You can ask it for what information might be missing. Everything about this is in the prompt and the questions you ask it.. Just asking "what is the diagnosis" leaves out a LOT of context, and what you expect.

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#227
One complaint he has is "It doesn't know to ask the right questions".

Well, the prompt was to give diagnoses, not questions. Ask GPT for the follow up questions first, then the diagnoses.

This is fascinating in that, because now the machine speaks human, we subconsciously ascribe human agency to it. This guy was instictively treating it like a colleague, who would naturally ask follow up questions unprompted. But you still have to prompt the machine properly.

So, 50% diagnosis success rate for the wrong prompt, for a LLM that can still grow, for a model that is not specialsed in medicine? In the literal first month of the "AI age"? Doctors are so done.

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#228

Earlier quoted context omitted.

A half working system is good enough for real life tasks when combined and used by and in conjunction with attentive half working human collaborators.

Translators who've been picking up after machine translation for the last decade will tell you that it is not, in fact, useful to them. They get paid half as much as before, but correcting the machine output is as much work as translating from scratch, and sometimes more, since the machine can give results that look good at first glance but are fundamentally incorrect.

I don't know much about professional translating, but wouldn't you be able to track the error rate of the machine translation by looking at the number of corrections? And then translators could use this as leverage to negotiate better pay if it's obvious they need are translating from scratch in, say, 50% of cases anyway.

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#229

One complaint he has is "It doesn't know to ask the right questions". Well, the prompt was to give diagnoses, not questions. Ask GPT for the follow up questions first, then the diagnoses. This is fascinating in that, because now the machine speaks human, we subconsciously ascribe human agency to it. This guy was instictively treating it like a colleague, who would naturally ask follow up questions unprompted. But you…

> Doctors are so done.

I would be curious to see the outcome if the patients entered the symptom descriptions themselves...

Re: I’m a doctor: Here’s what I found when I asked ChatGPT to diagnose my patients

#230

One complaint he has is "It doesn't know to ask the right questions". Well, the prompt was to give diagnoses, not questions. Ask GPT for the follow up questions first, then the diagnoses. This is fascinating in that, because now the machine speaks human, we subconsciously ascribe human agency to it. This guy was instictively treating it like a colleague, who would naturally ask follow up questions unprompted. But you…

Doctors are done? Can the chat bot take out a colon or gall bladder or fix an aneurysm? Hilarity.
Post reply on HN