Which medical specialties do you think are likely to be replaced by AI's in the next 50 years?

[Olle]

Recent news articles seem to suggest that AI's are becoming more accurate than trained physicians in medical diagnosis and surgery. 

There already seems to be a consensus that pharmacists will be replaced soon by AI machines.

Which medical fields do you think will be replaced within the next 50 years? Radiology?

[NeuroD]

48 minutes ago, Olle said:

Recent news articles seem to suggest that AI's are becoming more accurate than trained physicians in medical diagnosis and surgery. 

There already seems to be a consensus that pharmacists will be replaced soon by AI machines.

Which medical fields do you think will be replaced within the next 50 years? Radiology?

In 10 years maybe a couple like radiology/path/anesthesia....but in 50 years....all of them! Joking, but only a partly  :P.

[A-Stark]

Please state the nature of the medical emergency. 

[Eudaimonia]

Even though it would suck for our job prospects, I'd like to see a Baymax

[Edict]

I'm uncertain about complete replacement although it is a possibility, but I would wager that anesthesia, radiology, dermatology would be partially replaced. i.e. we wouldn't need as many of these physicians in each field.

Regardless, I think none of this would truly affect anyone of us. We would be realistically looking for jobs in about 5-10 years time and at that point I don't think much will have changed. Once we secure jobs and gain seniority, they would start off cutting numbers by reduced hiring rather than firing seniors. Also, do not underestimate the power of the medical establishment to delay change that is not in their best interest. Even if we had the technology in hand today, physicians would ask for studies and a gradual delayed rollout that could take a good 10-20 years. 

 

[A-Stark]

Not sure we'll have intubating machines or AI that's well suited to individual tailoring of anesthetic plans in the foreseeable future. 

[Eudaimonia]

For radiology, it seems like it's a competition between computers and pigeons 

http://www.sciencemag.org/news/2015/11/pigeons-spot-cancer-well-human-experts

[Hellothere77]

9 minutes ago, Eudaimonia said:

For radiology, it seems like it's a competition between computers and pigeons 

http://www.sciencemag.org/news/2015/11/pigeons-spot-cancer-well-human-experts

While adorable, that's about pathology. Just look at the image the pigeon is checking out

[Eudaimonia]

13 minutes ago, Hellothere77 said:

While adorable, that's about pathology. Just look at the image the pigeon is checking out

Oops, was easily mixed up. I'm not there yet haha these pigeons know more than I do 

[GrouchoMarx]

Pathology is the most in danger from AI as it is a particularly binary specialty. Cytology is specifically at risk of becoming entirely obsolete within a short time frame.

[Hellothere77]

6 minutes ago, Eudaimonia said:

Oops, was easily mixed up. I'm not there yet haha these pigeons know more than I do 

Well they're very smart pigeons

[ploughboy]

12 hours ago, Hellothere77 said:

While adorable, that's about pathology. Just look at the image the pigeon is checking out

We just need to train the pigeons not to poop all over the place.  That's a (partially) solved problem for pathologists.

[Hellothere77]

6 minutes ago, ploughboy said:

We just need to train the pigeons not to poop all over the place.  That's a (partially) solved problem for pathologists.

 

I don't know if it would fit a pathologist

[Snowmen]

I doubt any will be completely replaced since patients want their care to come from humans and not machine (at least that's what I prefer). Also, you could argue that with recent advances, imaging should have made some specialties obsolete since they can be used to diagnose many diseases almost single-handedly, yet that is hardly the case.

I think the most likely scenario is that AIs will be used as a "quality assurance" tool to double-check on the work done by human physicians or vice-versa.

[Koopatroopa]

25 minutes ago, Snowmen said:

I doubt any will be completely replaced since patients want their care to come from humans and not machine (at least that's what I prefer). Also, you could argue that with recent advances, imaging should have made some specialties obsolete since they can be used to diagnose many diseases almost single-handedly, yet that is hardly the case.

I think the most likely scenario is that AIs will be used as a "quality assurance" tool to double-check on the work done by human physicians or vice-versa.

Personally (and maybe naively) I'm excited for AI's integration, especially around summarizing current literature and best practices on a per patient basis. And, I agree that it will be a long time before it threatens to replace specialities. I hope it will make physician care safer and more efficient over the next half-decade. 

A (perhaps tangential) question: What tasks would we want to see AI take over?

I really hate paperwork! So I have a lot of personal enthusiasm for AI that can listen to and summarize patient meetings and care plans and give me more time to see patients in the day. Not sure how realistic this is, but I can dream, right?

[plastics91]

It will most definitely replace many tasks being done by any specialties that rely heavily on image based diagnosis (ex. path, rad, ophth & even derm to a degree). Detecting and categorizing image data is so advanced that it's harder for those specialists to argue they have added value for that specific task which can be bread & butter for most, not all specialists in that field. No specialties will be completely obsolete, and it can evolve around AI. It'll surely change compared to now though.

Frontiers of AI research is expanding at such fast rates it's difficult for anyone to give a valid estimate, but it's sooner than most people like to think and believe. Gov has high incentives to make this happen.. Don't expect them not to capitalize on it. Personally have ties that can speak to this direction; it's an inevitability.

On a side note, it'll be a complete turn around for what specialties will become lucrative, or maybe even just "safe". All bets are on primary care, especially family, whose scope of practice will expand to many areas as AI can replace narrow-scoped knowledge. Expanding FM+1 programs support this direction.

Any specialized or subspecialized areas are major targets for replacement. Generalist knowledge will become more valuable as AI is less proficient in that area. This could change of course but it's the safest route currently.

[Edict]

9 hours ago, plastics91 said:

It will most definitely replace many tasks being done by any specialties that rely heavily on image based diagnosis (ex. path, rad, ophth & even derm to a degree). Detecting and categorizing image data is so advanced that it's harder for those specialists to argue they have added value for that specific task which can be bread & butter for most, not all specialists in that field. No specialties will be completely obsolete, and it can evolve around AI. It'll surely change compared to now though.

Frontiers of AI research is expanding at such fast rates it's difficult for anyone to give a valid estimate, but it's sooner than most people like to think and believe. Gov has high incentives to make this happen.. Don't expect them not to capitalize on it. Personally have ties that can speak to this direction; it's an inevitability.

On a side note, it'll be a complete turn around for what specialties will become lucrative, or maybe even just "safe". All bets are on primary care, especially family, whose scope of practice will expand to many areas as AI can replace narrow-scoped knowledge. Expanding FM+1 programs support this direction.

Any specialized or subspecialized areas are major targets for replacement. Generalist knowledge will become more valuable as AI is less proficient in that area. This could change of i course but it's the safest route currently.

I can agree with some of what you said but i don't believe specialties will be less popular. I don't think any specialties will boom, but the specialties that I think will survive better will be the ones that interact with patients, so primary care would do well, internal medicine and subspecialties as well as surgery will continue to do well. 

Governments might want to replace specialists for cost reasons, but you underestimate the power of specialists to fight back. Ad campaigns saying "the government wants to replace your doctor with a robot... are you okay with that?" or "Unfortunately T3452 says that the $60,000 cost of surgery is not worth the 3.4% of survival, we cannot do this surgery... are you okay with a robot deciding whether your grandmother lives or dies?"

No government would want to wade into that minefield. 

People in the 60s were predicting that by now we would be in flying cars... I think there is a tendency to assume that change happens faster than it actually does, just like we have a tendency to catastrophize the future. I mean, look at our hospitals, most were built in the 60s, some even older, a lot of our hospitals are still using paper charts, the old guard never likes change and they slow things down. 

 

 

[plastics91]

8 hours ago, Edict said:

I can agree with some of what you said but i don't believe specialties will be less popular. I don't think any specialties will boom, but the specialties that I think will survive better will be the ones that interact with patients, so primary care would do well, internal medicine and subspecialties as well as surgery will continue to do well. 

Governments might want to replace specialists for cost reasons, but you underestimate the power of specialists to fight back. Ad campaigns saying "the government wants to replace your doctor with a robot... are you okay with that?" or "Unfortunately T3452 says that the $60,000 cost of surgery is not worth the 3.4% of survival, we cannot do this surgery... are you okay with a robot deciding whether your grandmother lives or dies?"

No government would want to wade into that minefield. 

People in the 60s were predicting that by now we would be in flying cars... I think there is a tendency to assume that change happens faster than it actually does, just like we have a tendency to catastrophize the future. I mean, look at our hospitals, most were built in the 60s, some even older, a lot of our hospitals are still using paper charts, the old guard never likes change and they slow things down. 

 

 

Yes I agree, I think it'll take awhile for popularity for a certain specialty to change because it's influenced by how current and future medical trainees perceive what's coming up. But the most likely way I think AI will manifest in medicine is: 1. Scope of practice for specialists will decrease as primary care will take over the most common cases, 2. Some level of primary care will be taken by NPs, PAs, but not to the degree of 1 bc general purpose intelligence is much harder to master for current AI than specific cases. I think everyone knows that in Canada jobs for specialties are already scarce, I project it will become scarcer. It'll still exist of course, but I think medical trainees will start to see more realistically what the odds are of becoming and what they may need to sacrifice.

As for robots replacing doctors.. if we're talking evidence based practices, as soon as robots become cost-effective, implementation is not only an option, but is the right one. Patients refusing robot doctors? Maybe generations before us, but after us? I don't think they'll be phased. Even for those older, healthcare resources are stretched to the max as it is, would it really matter if it means they can get the care they need? If we're talking about what's the best way forward to optimize outcomes for cost, I think there is good likelihood robots will be able to offer that better than humans for most tasks, though not all.

I think it's in the interests of medical graduates to read up on these issues, and get some basic math & comp sci training to better understand how things will shift, and how they can change their respective fields to maintain a value prop. It's not a mutually exclusive thing, AI and docs.. I think we can work together to improve healthcare in the long run. There may be growing pains, and some people won't be happy, but it's happened successfully many time. What's a society without progress  

[Edict]

In 50 years there may be changes like you described. Certainly though, no one reading this now should be concerned. Choose whatever specialty you want, you'll be looking for a job in 10 years or so and in 10 years nothing will have changed. 

[goleafsgochris]

I would not at all factor this into selecting a career.  Assuming a ~35 year career I don't think theres any real concern.  Maybe certain tasks can be automated within that time.  I suppose if I had to guess radiology could be most at risk?  But even then I would be concerned about outsourcing more than AI.  If you extend the timeline beyond 50 years of course its more up in the air 

[Lactic Folly]

Not particularly worried either. This figure from a recent article in CARJ outlines the role of radiologists in patient care. The areas where artificial intelligence are being applied are circled in red - as you can see, there are many areas that require human input to place automated output into the context of a specific patient's situation.

https://www.carjonline.org/article/S0846-5371(18)30030-5/fulltext

 

[path82]

Jobs are not lost; rather, roles are redefined; humans are displaced to tasks needing a human element. Radiologists and pathologists need not fear artificial intelligence but rather must adapt incrementally to artificial intelligence. JAMA. 2016 Dec 13;316(22):2353-2354

The question of human versus computer has now been refined as, human versus human with the computer. Arch Pathol Lab Med. 2017 May;141(5):622-623

Humans will remain essential to the intelligent use of AI in medical practice. Am J Med. 2018 Feb;131(2):129-133

[medigeek]

Replaced is not the right term. It's the very slow decrease of workload demand because everything is streamlined with AI. Suddenly the need for 10 doctors to do that workload is cut down to 8 doctors, then 6, then 3 etc. 

[cleanup]

On 4/19/2018 at 1:47 PM, Snowmen said:

I doubt any will be completely replaced since patients want their care to come from humans and not machine (at least that's what I prefer). Also, you could argue that with recent advances, imaging should have made some specialties obsolete since they can be used to diagnose many diseases almost single-handedly, yet that is hardly the case.

I think the most likely scenario is that AIs will be used as a "quality assurance" tool to double-check on the work done by human physicians or vice-versa.

You think that. But future generations will not. Even young folks born in the last 15 years or so have a grossly different view of technology than millenials and Gen X'ers and every generation beyond. The entire concept of AI even becoming a small part of our modern life was foreign just 20 short years ago. You couldn't get people to agree to self-driving cars. But the fervor couldn't be stronger. I trust my car to drive itself 90% of the time I'm on the highway. I couldn't have imagined that, nor could I trust it, when I was a child. People didn't even trust the Internet in the early to mid 90s. People didn't trust cars in the late 1800s. 

You cannot stop progress (and I mean this only in the sense of technological progress, not social progress). As Seth Godin says, every industry is funnelling inward to less and less human resources and human investment. This has been a continual push since the agricultural age. Just because our machines have gone from industrial to digital doesn't change anything. It only accelerates it.

Image processing, whether 2D or 3D, is very, very easily done by computers. IMO the only thing holding it back in the end are political & social barriers, but with time (and enough money from the right people), even those disappear.

You know those CAPTCHAs by Google where you have to click "All images that contain CARS" or "ROAD SIGNS" or "ROADS" or "STORE FRONTS"? All that is is Google collecting data to train self-driving cars. The same concept (and on top of that, machine learning) can be used to train computers how to identify, diagnose/interpret and report on almost anything. Not just medical imaging. It could be argued that imaging diagnosis is one of the simpler things to train a computer to do.

[Coldery]

On ‎3‎/‎10‎/‎2019 at 3:51 PM, cleanup said:

You think that. But future generations will not. Even young folks born in the last 15 years or so have a grossly different view of technology than millenials and Gen X'ers and every generation beyond. The entire concept of AI even becoming a small part of our modern life was foreign just 20 short years ago. You couldn't get people to agree to self-driving cars. But the fervor couldn't be stronger. I trust my car to drive itself 90% of the time I'm on the highway. I couldn't have imagined that, nor could I trust it, when I was a child. People didn't even trust the Internet in the early to mid 90s. People didn't trust cars in the late 1800s.  



You cannot stop progress (and I mean this only in the sense of technological progress, not social progress). As Seth Godin says, every industry is funnelling inward to less and less human resources and human investment. This has been a continual push since the agricultural age. Just because our machines have gone from industrial to digital doesn't change anything. It only accelerates it.

Image processing, whether 2D or 3D, is very, very easily done by computers. IMO the only thing holding it back in the end are political & social barriers, but with time (and enough money from the right people), even those disappear.

You know those CAPTCHAs by Google where you have to click "All images that contain CARS" or "ROAD SIGNS" or "ROADS" or "STORE FRONTS"? All that is is Google collecting data to train self-driving cars. The same concept (and on top of that, machine learning) can be used to train computers how to identify, diagnose/interpret and report on almost anything. Not just medical imaging. It could be argued that imaging diagnosis is one of the simpler things to train a computer to do.

I don't believe that technological progress is the main point of contention. I'm sure that technology will inevitably eclipse human beings in most possible faculties excluding those that require human ingenuity and creativity (?). 

 

The main argument is whether you can get human beings to personally accept handing over their ultimate fate to a non-sentient robot. 

 

From day one, the mass majority of us are inculcated with a certain sense of trust towards other humans having been raised by them and entrusting them with any concern we had about ourselves, regardless of the effectiveness of their help. 

 

By moving from a system with which we are all familiar to one that disregards it, we effectively relinquish:

1. Any sort of accountability that we used to have for others mistakes: A robot can't "pay"/feel any sort of true remorse for mistakes it makes. It's a robot.

2. Human-human interaction: One of the most vital characteristics of a good clinician is being able to show the other person that you actually care so they know that they aren't in this alone. This won't be the case with a robot doctor unless you: a) develop a robot that is indistinguishable from a human being (pass the Turing test) and b) don't let the patient know that their doctor is a robot => Ethical issues

 

There are very few, if any, technologies that are used in the mainstream which are fully operated by robots (no human control at all) that hold the balance of life and death. Airplanes, self-driving cars, and mass rail transit cars all still have human operators. Whether we will accept human-unaided AI tech for these parts of our life within the next several decades is somewhat unlikely. Second half of the 21st century or into the 22nd century? Maybe.

Science fiction loves to speculate about the future of everything we currently take for granted, a primary reason for its appeal. Whether things pan out the way they've speculated is another issue. Leonardo Da Vinci's blueprint for self-propelled flying contraptions weren't perfect predictions but there definitely were similarities. The only thing we can truly bet on is the inevitability of change. 

As for how this applies to the medical field, as most people have already said, it's unlikely that we are fully replaced by robots in the near future but AI as another check on diagnosis is most definitely a possibility.