Hacker News Reader

Comment by intricatedetail | Hacker News Reader

Some doctors use expert systems. They select symptoms and computer spits out possible list of treatments and then doctor picks one. If it doesn't work asks to come back and tries the next one. It's kind of like a human in today's self driving cars. Especially when it comes to mental health and anti-depressants. Essentially tests on production.

View on HN

This isn't even close to how doctors prescribe medications. You don't prescribe meds without having a working diagnosis. Once you have that, you use the knowledge gained in med school and residency to pick the first line drug. If there are contraindications due to comorbidities (which there often are), you have to figure out what other meds you can use. You can consult online resources (e.g. UpToDate) to look up second, third, fourth line meds as well as advice on specific complicated scenarios.

Trial and error with prescription drugs without a diagnosis as you suggest is malpractice. Maybe you're specifically referring to psychiatry? That specialty is uniquely difficult since our understanding of psychiatric diseases is still murky. But even within psychiatry there are best practice guidelines on how to manage and treat different diseases.

This is what I saw my doctors were doing. Also I saw cardiologist comparing my diagnosis using Google images.

Having software to show differential diagnoses, or using Google images because you know to search for, are not the smoking guns you think they are.

Yeah, I believe your experience, and I think it highlights the issues we have with communication in medicine. While it may look like your doctor is just blindly Googling something, I would imagine they're probably using it as more of a reference source (at least that's what I often do). I regularly use radiopaedia.org just to look up a quick fact or find alternative examples of a diagnosis I'm working with.

It's like Googling coding questions and reading a StackOverflow thread. Obviously no programmer is solely relying on StackOverflow to do their job as no physician is solely relying on Google, UpToDate, or any other resource. They're simply quick references.

> Obviously no programmer is solely relying on StackOverflow to do their job

I've encountered a few people who were doing something very close to this. I really hope that doesn't happen in medicine too.

That's down to the (obvious) fact that job performance and working ethics is not equal but distributed among practitioners.

I think this is especially visible in software "engineering" with people joining the craft after a few weeks of boot camp. (think engineering vs programming)

However, we put doctors through an especially rigorous and long training and certification process to minimize the amount of unqualified practitioners.

Doctors are supposed to have a higher entry barrier than software developers. Does not mean that all of them are brilliant, of course.

You might be right for single-casual illnesses like a broken leg, CoVid-19, Tuberculosis, and others. But it's a whole other thing when it comes to more complex illnesses.

Having been for 18 months through different types of psychiatrists and clinics, I came out quite surprised in how "trial and error" this whole system is.

I'm writing this from Switzerland, where we have an (arguably) high quality health care system. But the amount of "OK, that didn't work, lets try this other drug". Or, now, 18 months later, "Oh well, we did the list once through. But who tells me that the MD prescribing the first drug did a correct analysis? Lets start the list from top again." Or, for a friend of mine, his girlfriend found a working cure like that: "Oh, this brochure describes your symptoms so clearly, and it's completely different from what you've been treated for these last 20 years. Let's try it!"

Spoiler alert for my case with the list: the top of the list was not better the second time around.

If you think this is cynical, well, I would like at least _one_ of these drugs to work. If you think the MDs are all useless: well, at some times I was glad they were there.

So, well, I think having a little less than random system might be helping. Let the MDs watch if it makes sense, enter the correct diagnosis, and catch the stupid errors data entry people can make. But i'd give it a try...

The closest thing to an expert system I've seen is clinical practice guidelines, which sometimes includes decision trees of indications/contraindications for administering certain medications for common and time-sensitive medical events, like cardiac arrest or exacerbation of breathing difficulty with COPD.

Hey remember when the opioid companies paid the clinic management(?) software companies to push opioids to people?

https://www.washingtonpost.com/nation/2020/01/28/opioid-kick...

Reply on news.ycombinator.com