Much of this would be solved if you specified whether you used cash or accrual basis
346 karma · joined March 21, 2024
The medical world moves at a glacial pace compared to tech. Complaining that ML algos haven't swept the industry ignores all of the factors pushing it in that direction.
"Best medicine" operates on a consensus model of the most prudent decision-making given present knowledge and evidence. That takes time. Pushing boundaries as a doctor outside of a research environment doesn't earn you brownie points. It increases your chances of getting sued.
You also dramatically overestimate the amount of autonomy any given doctor has over the tools they are able to use. The vast majority are employees. It is like if you worked at Google as a SWE, and I came on HN ranting that you don't want to use the most recent release of Sonnet 3.5 to help you write features faster for Gmail, because you're "fiercely protective of your salary." You would laugh at how ignorant the complaint sounds.
Yep. It's like modifying an undocumented API where there are tens of thousands of endpoints and you know there are a few dozen services that rely on each endpoint, and probably a few dozen more that nobody knows about. You are only allowed test runs that are expected to succeed, since test failures = death. Your dev environment is a mouse. Every change gets pushed to prod with no option to revert.
> My appeal to you would be to allow more disruption in this space
No. There is far too much disruption that we don't want/need. If you want to help, go put yourself through medical school first, and then bring your CS knowledge to supplement it.
Jumping in when you haven't learned the basics wastes everyone's time like you have here.
Google paraneoplastic syndromes. Cancer is rarely a problem of just tumors.
> just "laser" all the tumours out and problem solved right?
No. The tumors don't just evaporate. The treatment causes tissue necrosis. Your laser causes inflammation. Even the high tech IMRT you're picturing has those problems.
>Metastatic cancer is defined by essentially there being too many things to hit though,
Metastasis as a concept is fairly meaningless in how you're using it. There are so many more factors in treatment of various cancers that have nothing to do with the distribution. There's a lot that clinpath looks at and frankly I care much more about what's on those reports than playing where's Waldo with nodules. Frequently we assume micrometastasis just based on the cell line or other characteristics.
> therein lies the entire detail
So no.
> you can't treat someone faster than they die of either the cancer or the treatment.
Again far more consideration goes into it but I'm losing my desire to debate these contrived simplistic takes.
Why?
Appealing to the authority of a pathologist is safe in my book given the subject matter and the education of the counterparty. There is a lower bound of requisite knowledge beneath which it's not worth even trying to entertain your claims about just "find and destroy rogue cells". You can't even begin to respond with logic and refutation if you haven't achieved anything close to the level of expertise required to be certified as a pathologist. Coming in here trying to play the angle of "well just model with supercomputers" isn't the winning ticket. Contrary to stereotypes of computer-illiterate doctors, there are some wicked smart people working in medicine (I've met and befriended many) and I guarantee your genius approach was considered and abandoned decades ago.
Im not sure who the target audience is for the article, but it strikes me as one where if the author were more knowledgeable on the subject, they would have many of the answers to what they are asking. It reads to me like someone writing "why are there still bugs in programming? Is it because of too many buttons on the screens?"
Im in agreement with you that most people don't grasp just how medicine as an area of study is unfathomably complex. It mimics a fractal in the sense that the deeper we go in trying to understand things it just opens up a wider scope of investigation.
Well that's a pretty wicked drawback if I've ever heard one.
This is another common misunderstanding of why people so commonly struggle to get a convincing diagnosis. The problem isn't a clueless doctor robotically listening for keywords. It's that the doctor has 10 minutes to dedicate to you and if it's not in the 80% of straightforward cases for them, the patient gets the bare minimum and then becomes someone else's problem. To solve that you need to fix provider shortages and (in the US) remove the corporate overlords constantly applying pressure on the little worker bees beneath them.
> out doesn't mean you can't look at the patients and see how things are turning out. And recognize that things are not static--your first patient doesn't exist. Things don't just sit there at three for three years and then kill you.
It's becoming clear to me you have a really odd sense of what doctors generally know and don't know. I don't really have an interest in exploring the left side of the Dunning kruger curve with respect to how medicine works.
In fact you are. You're speaking to one.
Your original statement completely ignores the massive amount of barriers put up during your training. From my own experience I can tell you that students are usually quite self aware that they are learning and on a short leash. The most egregious mistakes I've seen were almost always a result of understaffing and gaps stemming from leadership issues coming from the top down. The small mistakes and cut corners are from tired overworked people.
Nobody accepts "hurting people" as a component of training. It's an absurd statement.
I've found that usually the people who said that to me were the least prepared for what that meant.
This is overly simplistic. There is little to debate about your example of living at a 7/8 for 2 months. What is more realistic is some hypothetical situation where the average patient has an expected survival of 6 to 18 months, with the hope that the 7/8 pain can be reduced to a 4/5, and with acknowledgement that some days may be a score of 5, some days may be a score of 9. Then what?
As a provider on the other side of it you witness two similar patients take two different paths. Perhaps the first patient lives 36 months with pain score averaging a 3. The family will often appreciate the intervention and effort. The other patient experiences a string of complications and dies after 2 weeks, living at a pain score of 10. Their family paints you as an evil torturer who should have never been listened to.
In fact you can see both types of opinions in this very thread.
Modern medical education doesn't work this way.
> Xenon is considered “the perfect anesthetic” for 5 reasons:
> It’s extremely fast-acting — xenon gas kicks into full effect within seconds
> It wears off quickly — once the gas is removed, normal consciousness comes back within a minute or two
> It doesn’t interact with other drugs
> It leaves no lasting side effects or toxicity
> The anesthetic state xenon induces is extraordinarily powerful
The vast majority of drug incompatibilities are not a result of the physical interactions between the molecules. It's the undesirable effects that they exert in tandem on physiology. A vasodilator and a negative inotrope don't have to chemically react with each other to give a patient a very bad time.
I will acknowledge that xenon does appear to have interesting effects, but nobody should be base their health and safety on what the author has written. Their analysis on its safety is far far far too superficial to be worth anything.
Even that is a faulty system.
I really don't understand how people can think Tesla and Waymo are anywhere close to each other. Making a car drive itself is a much harder problem to solve than scaling production. If you look at what Tesla has, today, it is a vehicle that can sometimes drive on its own with the need for a human behind the wheel. The ability to remove the human from the front seat is the other 95% of the work. In systems like this where human lives are at stake, getting "mostly there" with FSD is essentially worthless.
The problems of logistics and scaling are well-known and well traveled. The invention of a self-driving car is not. Once Waymo has solved the issue of entering new cities and quickly gathering data to feed into their model, they've won.
Granted in my work I've never made the attempt to smell fentanyl, but it isn't one with a reputation of having an odor. I assume you mean the smell of recreational users of fentanyl.
If you start typing your assets and debt into the inputs the page width increases.
It tells you it is chat gpt 3.
Im not sure what would be the reason for paying 10/mo for this when everyone has gpt 4o....