Medicine automation

How sophisticated a machine has to be to replace:
General practitioner
Radiologist
Dentist
Surgeon?

How far are we from robodoctors?

Other urls found in this thread:

en.wikipedia.org/wiki/Tricorder_X_Prize
tricorder.xprize.org/
twitter.com/SFWRedditVideos

I thought we already had machines that can google for directions.

just a small bump

Probably not in our lifetime.

Surgeons won't be replaced by robots, they'll be replaced by regenerative medicine. Stem cells are key to this but they still have huge problems with heterogeneity, tumorigenicity, controlled differentiation, etc. They still can't even deliver them properly to target locations, they go all over the body.

Radiologists are being replaced...by doctors in India who read film for you, but hopefully President Trump will fix that. In terms of machine learning / computer detection, it's pretty far away. Some radiology tests produce a ton of data, and it's very difficult to do it for a general population.

GPs will probably be replaced by NPs or PAs. They cost less and less doctors want to become a GP for that low of a salary. (why make 160k a year after 12 years training and work crazy hours when you can be a PA or NP in two years and make 100 - 120k a year and work basically 9 to 5. )

Trouble with that is somebody's gotta be responsible for the care they're providing and you bet your ass some phone app you installed won't give a shit.

So what will happen is these machines won't replace doctors, they'll make their lives much easier and richer.

So far into the future if never. Medical cases are rarely cookie cutter and current machines can't even reliably read basic x rays, let alone be able to squeeze out and manipulate information from people who on average will read at a 6th grade level. Not to mention that medicine is socially one of the most protected and celebrated careers of modern times so politics will protect doctors even after they could be replaced with machines. Lastly people in today's society would never accept a diagnosis from a machine and would demand second opinions from humans.

If you look up how likely which jobs are to be automated in the future, medical care is always placed as one of the last.

real, humanlike AI

otherwise - never

We already have programs that can diagnose certain illnesses more accurately than humans. 10-20 years for it to be refined and go mainstream.

That's sad. Is at least lowering intelligence, wealth and skill threshold possible? Training analytic engine operators and anamnesis collectors in 2 years instead of qualified doctor in 7 or more?

I suppose that if patient won't know that his diagnosis has been made by a machine, he will accept his treatment. Responsibility will be on the machine operator or owner.

Is it possible to reform medicine the way goods manufacturing was reformed?

Who would benefit from this? Again medicine is highly revered and protected, no one will profit from this, not the patients and not the doctors only the machine manufactures. Treatment may cost less, and that's iffy, but medical research and overall medical care quality will drop significantly. So I doubt such an idea would ever be allowed to go through and I doubt anyone would to be treated by such people. Most patients don't like being treated by PAs and RNs as is.

I think surgeons are most likely to be replaced first. Why hire someone to cut a gaping holes in you and only be precise as human hands can be when you can have a machine do it through a 1/4cm hole for only the cost of electricity and be orders of magnitude more precise? Human surgeons demand more pay and are a greater liability.

Robotic surgery is already a thing. The robot is operated by a surgeon tho.

Surgery is too complicated for it to be completely replaced by AI anytime soon.

I thought quality will increase. I'm yet far from medical practice, but I believe computer could outperform a human in optimizing the treatment for individual needs. Precise calculation of drug doses, diagnosis and choice of prescribed based on latest evidence and statistics. And maybe much faster, because there is limited number of doctors and as a result queues.
If a net of such machines becomes big enough, it could adjust itself in real time.(in terms of effectiveness or ineffectiveness of certain measures)

I've heard that doctors in Europe follow clinical protocols anyway.

>Who would benefit from this?
Areas that have half a doctor/ten thousand would benefit.
All they would need is a few servers.

The problem is when you need to treat people.

OP here
In OP I listed specialties in the order of anticipated difficulty of replacing. (IMO)

Who do you think will be replaced first the doctors or the engineers.

This. Urology is all robotic, but other specialties it's not worth the investment when outcomes are similar using MiS.

>Hello Patient #2569794
>My algorithms show the most probable diagnosis to be: Cancer
>'n-no i just had some really bad food and a headache'
>You will receive full-body radiation now. Please stand still.

You underestimate the human element and you massively overestimate computers. Also doctors will always be needed outside of a strictly clinical environment. Tough luck OP.

Medical AI.... are you nuts? And we will let big pharma program it for us to so no cure to any ailment is found and we are stuck in a cycle of sickness and system treatment. Think!!!!!!!!

Engineers.
Internal medicine will remain since having a doctor with human interpersonal skills is very comforting and people probably won't feel safe being "taken care of by WebMD".

Engineering however... AI guided drafting and design would be a very significant advance that has no such interpersonal restrictions (or at least, not as many).

Every time someone has made the argument that new technology won't catch on because of societal reasons, they have been wrong. We'll start seeing machines replace medical professionals in the next ten years---the technology already exists, it just hasn't been perfected---and will likely see human doctors reduced to the status of a novelty in about 50 years. Efficiency, and cost effectiveness always beats "the human element."

>Who would benefit from this?
Everyone. There's obviously a profit to be made by the manufacturers and healthcare providers---that is, insurance and the people who actually provide access to the machinery---but there's also a benefit to patients. They're no longer being examined by a person who has dozens of other patients to attend to, is prone to forgetfulness and bias, and can't possibly check for every scenario. There's too many advantages and almost no drawbacks.

as a doctor, I can tell you that a majority of people going to the doctors are people looking for empathy, they want someone to give a shit about their problems and help them out. There is a certain portion that probably wouldn't give a shit dealing with a machine, that's true, but depending on the time and their affection that might also change. You are under a lot of pressure, you made some mistakes, you sleep like shit, you're fighting with your wife, what's a machine going to do ? They spout out a diagnostic and prescribe pills ? some doctors do that, they're generally pretty unsuccessful.

If by doctors you mean therapists, I'm pretty sure constructing a program capable of empathy indistinguishable from humans is trivial.

>Is at least lowering intelligence, wealth and skill threshold possible?

God I hope so. The shit I have to memorize at school is the kind of stuff computers excel at. I'm talking about objective lab test data => treatment look up tables. They're backed by evidence.

>Who would benefit from this?

Well pretty much everyone will benefit from medical technology. The thing is all healthcare software will come with a big ass label that says:

>this app doesn't substitute medical advice
>consult an expert if you have questions

You'll be able to use them but if things blow up in your face you can't hold them accountable. You'll just get told you should've seen a real doctor.

I'm not so sure about that. With current technology, robots cannot replace surgeons. They can only be valuable tools.

A few weeks ago I witnessed a case where the surgeon transfixed a major artery during suture. This is one of the reasons people stay in the hospital for a few days after major surgery. Surgeons sometimes make serious mistakes like that. They're human. Nobody likes to admit it but studies on the matter prove that doctors make mistakes just like everyone else.

However, they have the capability to deal with acute complications which arise during the operation. The list of possible complications is huge. Managing them is complex, they show up while you're in the middle of something else and present in unique manner.

Before you can even begin to think about replacing human surgeons, you need to account for complications. It's hard enough to get the robot to perfectly perform an operation on a human being... And that's just a fraction of the task: it also must be able to perfectly deal with a huge list of potential complications which can happen in the middle of the operation.

Everything people know about industrial process automation doesn't quite apply here. In that field, every single variable can be accounted for; you can guarantee that the system won't fail. When a robot moves in order to perform some operation on some part, it is guaranteed that the part will be there. Even then, errors happen and parts are lost; there's a probability of that happening and you know how much money you lose, and you know whether it's profitable or not.

With people none of that applies. Every person's different in some form. Anatomy is irregular and it can vary naturally, it can be distorted by injuries and trauma, it will likely be obscured by liters of blood during the operation. Life-threatening complications can happen in the middle of the operation and you have to stop everything and deal with them. You cant afford to fuck it up

TL;DR it's not easy.

All the posters in this thread are missing the point that you don't need total replacement with automation for there to still be huge job losses. Even with partial automation you can have one doctor doing what used to take five doctors. That will still lead to major unemployment.

>I've heard that doctors in Europe follow clinical protocols anyway.

That's what doctors do nowadays. Evidence-based medicine is statistics-based medicine. Prevalent problems like diabetes have so much research that it's already pretty clear which treatments are most effective given objective patient data. This large body of research gave rise to guidelines, which are literal manuals about the disease and how to treat it. The manuals are filled with algorithms, flowcharts. Doctors simply execute them. Of course, they take the patient's individual characteristics into account.

People in this thread are the living embodiment of the stereotype of the moronic computer science undergrad who goes "I'M GOING TO SOLVE EVERYTHING WITH ALGORITHMS".

>In terms of machine learning / computer detection, it's pretty far away


Oh boy are you in for a surprise

>How sophisticated a machine has to be to replace

as sophisticated as human brain is

>In terms of machine learning / computer detection
Where do I learn more about this? Specifically with its application in radiology?

based on the doctors Ive seen over the years, I could get better results from google and a pharmacy without a security system

Case in point: I'm not sure you're a human or a robot.

You should work on your own empathy programming, gearhead.

Interested in this topic as well

bump

>heres an anecdote showing a fictional situation meant to scare you
>point proven

it's like you're not even trying

not sure about radiology, but you might find this interesting.

looks like they picked winners.
en.wikipedia.org/wiki/Tricorder_X_Prize
tricorder.xprize.org/

We can take this further.

>Hello patient 15706, it appears you have clinical autism
>O-oh...can I at least say goodbye to my parents?
>I am sorry, off to the Veeky Forums maximum security prison for you.

Even if we'd get machines sophisticated enough to perform the complex and very human work of good doctors, I doubt it would immediately be adopted as an alternative and put doctors out of work for two reasons. The first being that doctors are pretty clever, if something like machine uprising in the workplace was to threaten their jobs I'm convinced they'd figure someway to slow down to process or even stop it.
Second reason being that a lot of people still harvest a very deep fear or machines and robots, therefore it would be difficult for the population to accept it and we'd probably see a small rise in alternatives medicines, slowing down the process.

You cannot teach a machine how to process patient centered medicine and make difficult division that deviate from things programmed into it "from the book. not only that, often there are clues given to doctors by patients that only experienced human doctors can pick up on to make a diagnosis be them general practitioner, or cardiologist, (lol surgeon or dentist). Robots are only as smart as they are programmed to be and that is not something we can rely on in medicine.

See Fucking this

When I shadowed a neurosurgeon, an NP just walked out of the OR at 5 because she wanted to go home kek

The only reason I went with engineering was so that I can contribute to making medics obsolete and homeless. I just fucking hate medfags so fucking much.

Unless you want to do something very specifc that you have to have a MD to do it is 1000x better to be a NP or PA for many reasons, including the one you just said.

You can teach a machine to teach itself. The human brain is just a machine, programmed by billions of years of evolution. Why couldn't a machine designed to teach itself, manage to teach itself?

>This is what computer science fags actually believe.

Sure when we finally even begin to work out consciousness and the mechanism of incorporation of novel experiences, then maybe we can have a semi serious chat.

The funny thing is engineers would probably be more easily replaced with robots than doctors (considering most of you already are lol) not but really you are being vary naive about the power of AI.
I'm not saying robots won't be able to make medicine more efficient(they already are in some ways with surgeries) but completely replace? Not by a long shot.
You will need human level of adaptability, emotional, and traditional intelligence for those robots to be effective. Humans are not all the same like a factory machine or pieces in an assembly for a final product. Even if you updated these robots periodically to deal with these novel experiences, it would have to be on the fly, especially if we're talking about surgery or any procedure heavy fields.
Gets it.

There will be a certain amount of remote conference types of situations in which doctors manage big groups of mid-level providers and nurses without actually seeing the patients, or seeing them via phone/teleconference (which is already pretty widespread).

The problem is that a lot of times in medicine there isn't a really definitive "this is what the absolute best thing is to do," especially when there are things like individual patient values, lifestyles, genetics, etc. at play. As a patient, you are choosing your doctor and trusting him to make recommendations on your behalf. If everything is theoretically guided by AI, it'll have to be completely 100% standardized, and now you lose that ability to get individualized care that comes along with personally knowing and trusting your doctor.

Also, a lot of what makes a doctor a good doctor is intuition and wisdom, which can't be programmed. Someone fresh out of medical school can't practice independent medicine for that very reason, they haven't seen enough cases. It doesn't matter how much you've memorized the textbooks, read the flowcharts, and studied the terminology. Good doctoring is about experience.

Not to mention the fact that I'm not sure people -want- robots being their caretakers. People are already very critical and hard to please in terms of their doctors practicing compassionate medicine.

Plus the liability would be a nightmare. With human providers, doctors take a huge brunt when it comes to personal liability for anything from medication side effects to personnel mistakes. With independent robotic physicians, the costs would be astronomical.

And also what that other guy said about accounting for complications and reacting to them, especially in surgical situations.

>However, they have the capability to deal with acute complications which arise during the operation. The list of possible complications is huge. Managing them is complex, they show up while you're in the middle of something else and present in unique manner.
This is a prime argument FOR AI doctors, though. Computers exist for the sole purpose of managing huge, complex tasks. I read somewhere that as much as 20% of outpatient cases are misdiagnosed. Machines don't have to be 100% perfect; to replace you, they only need to be better than 80%.

>General practitioner
won't be replaced by a single machine, but by various diagnosis systems. I expect in the future a doctors visit will do an on the spot blood test and DNA analysis and check everything going on with you. A doctor will only be needed if things are out of range
>Radiologist
This will be automatable in the next 5 years. Very easy computer vision problem
>Dentist
Dentistry is a fucking hack profession, and only exists because dentists are slow to modernize. 99% of modern dental treatments should probably go away, but the ADA won't do anything because that would mean destroying their entire industry.

Regrowing teeth, and treating cavities without drilling are all possible TODAY, but modern dentistry won't adopt them because they are stuck in their old ways.

>Surgeon
Machines will eventually do the cutting, I say 10 years on this. But I think humans will be in the loop for a long long while.

Its possible for specific types of surgeries, a machine could be made which could be run by a technician instead of a surgeon.

Radiologists are already being outperformed by neural network based classifiers, but you still need experts to read the output, relay it to the patient / other doctors, decide what to do next etc

Neural nets can do 90% of the work of GP's as well

Surgeons do make use of robotics to aid surgeries in some cases, but full automation of surgery is a long way off, though it's definitely possible

Almost every subfield of medicine is using machine learning. Radiology, hematology, cardiology; basically any field where a set of known parameters are used to diagnose or make predictions

>political sway
True

>medical care being last to replace
Similarly to how we'll want real life waiters at fancy resturants. Reanimating dead boddies would do the same thing.

>rest of post
Total garbage. Respect and fear linear algebra.

>so politics will protect doctors
You really think doctors have any political clout compared to insurance, tech, and the actual people who employ said doctors?

You really think technicians are going to take the job of surgeons? Surgeons will learn the skills of technicians and will absolutely not give up their jobs and further more people wouldn't trust technicians instead of surgeons for an operation.

>treating cavities without drilling

tell me more user, i'm supposed to get a filling soon.

NP and PAs do the bitch work.

Your surgeon will never wipe a homeless persons ass. Your surgeon will not give prepare meals for patients.

Also Neurosurgeons make a little less than 1mil a year. I'd rather spend 1 1/2 decades of training to make that salary than to become a plebian nurse.

I'm making the point that the algorithms we have today come to the wrong solutions based on statistics that don't factor the element of the individual. I'm also making the point that humans desire empathy and attention. Stop being obtuse.

Thats not what I said. I think surgeons will stay in the loop for a long while.

>this thread is still going on
>comp-sci freshmen are still embarassing themselves with their predictions and hypotheses
Why is every single person actually involved in the field far more pessimistic on the matter than you guys?

> interpersonal skills can't be touched by automation!

sure they can't

Computers are already better at diagnosing that doctors. The pharmacy can be ran by computers. Robots already perform surgery. We are not that far until you look at cost. Cost of those robots is the only thing saving doctors. You will replace doctors before nurses and white collar jobs before blue. Robots have a hard time building a sky scraper on their own or maintaining oil rigs in the gulf. They don't have a hard job sitting in a room and diagnosing patients ailments.

>Computers are already better at diagnosing that doctors
No, computers are better at certain diagnoses in certain fields after being fed data that has already been rendered comprehensible to them. It's a completely different thing.
>Robots already perform surgery.
They're not automated.

>you can't see the future implications of these things that already exist
>you don't understand technology advancement
Stay pleb.

>everyone is going to have a flying car by the year 2000!
Dude, you're the pleb. You're basically like those I fucking love science guys.

Not even close to the same. If we had some cars that could fly in 1900 then that would be an apt analogy. Also, the question is when it will happen. And only an absolute moron would literally look at real world example of it happening and say it can't happen. Flying cars COULD happen, but there is no advantage so why waste the time and money? Replacing over paid doctors with super reliable tech WILL happen and IS happening. Are you literally a retard?