Residency and obsolescence

[Endoxifen]

---

Members do not see this ad.

Considering how quickly medical technology has been moving in the past 16 years, should medical students be considering the possibility of their specialty being replaced, when picking a residency? If so, which specialties are more susceptible to this form of obsolescence? Radiology and anesthesiology come to mind, but what about radiation oncology?

 

[NecrotizingFasciitis]

This is an interesting question, and part of the reason I'm commenting is to follow this thread. But also, I've heard cardiothoracic surgery is at risk of becoming obsolete.

 

[NotASerialKiller]

I've heard cardiothoracic surgery is at risk of becoming obsolete.

I'm pretty sure most their jobs have already been stolen by interventionists

 

[marstonne]

This thread pops up constantly in pre-allo.

I was once concerned about this as well. And while I am still very much wet behind the ears, even as an MS-3 I can see that virtually no field is in danger of becoming outright obsolete. Scope of practice and demand for certain specialties will change, that is a given. But medicine is so complex and not nearly as advanced as at least I thought it was. Our understand and treatment options for many common illnesses is so rudimentary, and there is so much work to be done in essentially every area of medicine. Keep an open mind and pick a field that has pathology and patients you enjoy working with. Those things will not change. The day to day of a specialty very well might, and you will have zero control over that.

CT surgery is in no danger of disappearing. I certainly hope a CT surgeon is around if I ever have a type A dissection.

 

[7331poas]

https://www.mskcc.org/about/innovative-collaborations/watson-oncology

 

[Endoxifen]

https://www.mskcc.org/about/innovative-collaborations/watson-oncology

That's exactly it though. I don't expect any significant changes during my education, but by the middle of my career, I don't know what it will look like.

 

[redferrari]

In the '50s, everyone thought we would have self-navigating, flying cars. They also predicted our meals being in a pill form. Biggest eye-opener for me thus far in my first year is how much we still don't know, and how we laugh at tinctures and potions homeopaths from 1880, but in reality that's how we will be remembered and the "quacks" from the old west were actually on to a lot of stuff.

 

[Endoxifen]

In the '50s, everyone thought we would have self-navigating, flying cars. They also predicted our meals being in a pill form. Biggest eye-opener for me thus far in my first year is how much we still don't know, and how we laugh at tinctures and potions homeopaths from 1880, but in reality that's how we will be remembered and the "quacks" from the old west were actually on to a lot of stuff.

I have no illusions about what we do and do not know. I do cancer research and I can tell you that we know next to nothing. Still, that means nothing when we are talking about non-research positions. Radiologists are essentially looking for patterns. Though that can be said for all analytical fields, I think that some require less day-to-day insight than others. That being said, I very well could be concerned over nothing.

 

[FindersFee5]

I have no illusions about what we do and do not know. I do cancer research and I can tell you that we know next to nothing. Still, that means nothing when we are talking about non-research positions. Radiologists are essentially looking for patterns. Though that can be said for all analytical fields, I think that some require less day-to-day insight than others. That being said, I very well could be concerned over nothing.

I've thought about that a lot, especially when I talk to my brother about pattern recognition and neural network learning. Computers are getting really, really good at it: think about facial recognition software, or speech-to-text, or handwriting analysis. Computers aren't programmed for these, they're just given enormous sample sizes and self-teach within programmed parameters (essentially). I think imaging reading is just waiting for somebody at google/amazon/etc to realize this inefficiency in the medical market and swoop in, especially with the rapid growth of EMRs and everything being electronically stored. Obviously, it isn't going to get rid of radiologists entirely, but I think it is going to drastically decrease the need for them.

 

[Endoxifen]

I've thought about that a lot, especially when I talk to my brother about pattern recognition and neural network learning. Computers are getting really, really good at it: think about facial recognition software, or speech-to-text, or handwriting analysis. Computers aren't programmed for these, they're just given enormous sample sizes and self-teach within programmed parameters (essentially). I think imaging reading is just waiting for somebody at google/amazon/etc to realize this inefficiency in the medical market and swoop in, especially with the rapid growth of EMRs and everything being electronically stored. Obviously, it isn't going to get rid of radiologists entirely, but I think it is going to drastically decrease the need for them.

In the short term certainly, but by the end of our career, it is conceivable that there will be no new radiologists. My physiology professor told us stories about how a friend of his, during residency, would check all of the EKGs for accuracy. Now, that job doesn't seem to exist anymore. I could be wrong, but if it's true, I don't see why that wouldn't extend to radiology or anesthesiology.

 

[FindersFee5]

In the short term certainly, but by the end of our career, it is conceivable that there will be no new radiologists. My physiology professor told us stories about how a friend of his, during residency, would check all of the EKGs for accuracy. Now, that job doesn't seem to exist anymore. I could be wrong, but if it's true, I don't see why that wouldn't extend to radiology or anesthesiology.

I honestly don't think anesthesiology is in trouble for a while. Yeah, a lot of it is reading graphs and pattern recognition, but there are also a lot more procedural aspects than many people think of (airway management, injections, post-procedural care, etc.)

 

[nlax30]

It's an interesting question but to be practical, I just don't see a specific field/specialty going away or being replaced by a computer in the near future, at least in our generation.

Obviously AI/machine learning is getting better and I do see it getting more and more important in various fields of medicine (imaging analysis, cancer therapy/genetics analysis, etc...) though ultimately at some point we're still going to want some sort of personal interaction to interpret those results with a patient. It may get to the point when the technology gets to that point before society or the law is.

There's currently a lot of talk of autonomous driving with a lot of folks thinking that the technology will be ready before society in general or the law is ready to allow fully autonomous vehicles to operate. Allowing a similar transition to happen in medicine is much more complex and involves more variables in my mind so is going to take much longer.

 

[Endoxifen]

It's an interesting question but to be practical, I just don't see a specific field/specialty going away or being replaced by a computer in the near future, at least in our generation.

Obviously AI/machine learning is getting better and I do see it getting more and more important in various fields of medicine (imaging analysis, cancer therapy/genetics analysis, etc...) though ultimately at some point we're still going to want some sort of personal interaction to interpret those results with a patient. It may get to the point when the technology gets to that point before society or the law is.

There's currently a lot of talk of autonomous driving with a lot of folks thinking that the technology will be ready before society in general or the law is ready to allow fully autonomous vehicles to operate. Allowing a similar transition to happen in medicine is much more complex and involves more variables in my mind so is going to take much longer.

I can appreciate that. Still, with economics as a driving force, I can imagine the politics moving much faster than we expect.

 

[HuskyMD Emeritus]

Reading x-rays and CT scans and pattern recognition may be programmed and done by a computer, but putting the images into clinical context is what YOU as the doctor have to do. I often review the x-rays and CT scans with the radiologists for their insights. You can't do this with a computer.

 

[marstonne]

Reading x-rays and CT scans and pattern recognition may be programmed and done by a computer, but putting the images into clinical context is what YOU as the doctor have to do. I often review the x-rays and CT scans with the radiologists for their insights. You can't do this with a computer.

I'm on surgery right now and we review scans/discuss cases with the radiologists almost every day.

 

[xffan624]

I've thought about that a lot, especially when I talk to my brother about pattern recognition and neural network learning. Computers are getting really, really good at it: think about facial recognition software, or speech-to-text, or handwriting analysis. Computers aren't programmed for these, they're just given enormous sample sizes and self-teach within programmed parameters (essentially). I think imaging reading is just waiting for somebody at google/amazon/etc to realize this inefficiency in the medical market and swoop in, especially with the rapid growth of EMRs and everything being electronically stored. Obviously, it isn't going to get rid of radiologists entirely, but I think it is going to drastically decrease the need for them.

If a computer can't even read an EKG (which is only a 2D simple image) properly half the time, I really doubt it's going to replacing a radiologist in reading chest x-rays any time soon.

 

[LizzyM]

In the early 1980s, I met several elderly physicians who had to retool after their careers as pulmonary specialists caring for patients with polio and TB dried up in the 1950s. You just never know. On the other hand, in the 1970s we could not have predicted the effect that HIV would have on internal medicine and specifically on infectious disease (first as a horrifying, incurable communicable disease and then as a chronic, managable condition).

 

[Espressso]

I dig this thread.

But in my opinion... We're not at the point where specific fields will be completely eliminated with advancements in technology. Yes, self-driving cars will take away the entire driving sector in the economy in our lifetime. But I don't think you're going to see an entire specialty eliminated because of technology. Even with Tesla's autopilot feature (which is so crazy cool when you ride in one)... humans are still instructed to keep their hands on the wheel. Similarly, even with the fanciest and most cutting edge surgical robots cutting open a knee, you will still expect (and require) human eyes, hands, ears, paying close attention during the process.

 

[Espressso]

Either way though, tech geeks like myself can't help but look at the changing technological landscape as it pertains to medicine and be super excited

 

[aldol16]

I think radiation oncology specifically might have some basis for concern because we've been treating cancers with radiation for the better part of the 20th century and the field itself is maturing. Initially, it was founded upon a philosophy of "blast it with all you've got and maybe you'll kill the cancer cells before normal cells." Now, technology has obviously gotten more specific and selective (I'm by no means an expert in the field), but it's still based on that sort of philosophy, which makes it vulnerable to more targeted approaches that may cause fewer side effects. Now, nobody can predict whether somebody will discover a more targeted approach to cancer therapy in the next 20 or 30 years, but many people are working on it on the research and development side.

 

[Lucca]

It's worth thinking about, though nobody can predict what the future will bring obviously.

I think the things to look for are:
1) What problems are in the process of being solved by new technology?
2) What fields are those problems in?
3) Are there structural pressures to pushing physicians out of those fields given those problems are solved by technology? (I.e. is the new technology cutting costs by excluding physicians? Is it safer than a human? Is the technology less invasive or safer than existing interventions? Does the technology require a more specialized physician or simply a skilled technician?)

Edit: it's also worth noting that fields are born out of technological advancements, not just destroyed by them. Neonatology is only a thing because we now have the technology to intervene and actually help "the neonates". Cardiothoracic surgery is losing procedures to interventional cardiologists...but there are also more interventional cardiologists.

 

[7331poas]

In the '50s, everyone thought we would have self-navigating, flying cars. They also predicted our meals being in a pill form. Biggest eye-opener for me thus far in my first year is how much we still don't know, and how we laugh at tinctures and potions homeopaths from 1880, but in reality that's how we will be remembered and the "quacks" from the old west were actually on to a lot of stuff.

And what do you say to machine workers in the US as well as truck drivers in the US?

These changes are real. We have machine learning. I am certain by the time I am dead that radiology/pathology will be a completely changed field, if it exists at all.

 

[7331poas]

If a computer can't even read an EKG (which is only a 2D simple image) properly half the time, I really doubt it's going to replacing a radiologist in reading chest x-rays any time soon.

https://med.stanford.edu/news/all-n...gists-in-predicting-lung-cancer-severity.html

http://www.nature.com/articles/ncomms12474

 

[jqueb29]

As an MS3, I'm not worried about this whatsoever. It's easy for the layman or even the pre-clinical med student to not know any better and worry about this happening. Once you actually get into the clinics and hospitals, you'll see for yourself that it's an absolute joke to even think that computers could come close to replacing any specialty.

 

[7331poas]

As an MS3, I'm not worried about this whatsoever. It's easy for the layman or even the pre-clinical med student to not know any better and worry about this happening. Once you actually get into the clinics and hospitals, you'll see for yourself that it's an absolute joke to even think that computers could come close to replacing any specialty.

http://www.nature.com/articles/ncomms12474

Yet we already have research efforts to replace radiology and pathology...

 

[WedgeDawg]

http://www.nature.com/articles/ncomms12474

Yet we already have research efforts to replace radiology and pathology...

The article is about using histological features to predict outcome, not make a diagnosis (though they show that the technique can distinguish between SCC and AC; however, they don't explicitly compare this to the ability of a pathologist to do the same, and the diagnosis of NSCLC had to already have been made before attempting to distinguish between SCC and AC). They used slides from a database that had already been given diagnoses. This is exactly the same thing as using something like the APACHE II score to predict mortality in ICU patients. Prognostication by nature is a data-driven endeavor and not something that humans can do by themselves. Using specific features of slides to predict outcome is not a "research effort to replace pathology". It's a tool to help pathologists increase their ability to help patients.

 

[Nommonnom]

I've thought about that a lot, especially when I talk to my brother about pattern recognition and neural network learning. Computers are getting really, really good at it: think about facial recognition software, or speech-to-text, or handwriting analysis. Computers aren't programmed for these, they're just given enormous sample sizes and self-teach within programmed parameters (essentially). I think imaging reading is just waiting for somebody at google/amazon/etc to realize this inefficiency in the medical market and swoop in, especially with the rapid growth of EMRs and everything being electronically stored. Obviously, it isn't going to get rid of radiologists entirely, but I think it is going to drastically decrease the need for them.

Ive read articles on a few startups that have already done this and claim higher accuracy of diagnosis than specialists. They were blocked from complete replacement of the specialty fields by lobbying groups representing the specialists. They now focus on making tools to be used by the specialists.

 

[7331poas]

The article is about using histological features to predict outcome, not make a diagnosis (though they show that the technique can distinguish between SCC and AC; however, they don't explicitly compare this to the ability of a pathologist to do the same, and the diagnosis of NSCLC had to already have been made before attempting to distinguish between SCC and AC). They used slides from a database that had already been given diagnoses. This is exactly the same thing as using something like the APACHE II score to predict mortality in ICU patients. Prognostication by nature is a data-driven endeavor and not something that humans can do by themselves. Using specific features of slides to predict outcome is not a "research effort to replace pathology". It's a tool to help pathologists increase their ability to help patients.

It is right now. But in 2080 thats not where this will be. Its drop dead naive to think that humans are better than computers in the long run at....anything.

Watson is already doing these type of things at Sloan kettering.

 

[wizzed101]

It is right now. But in 2080 thats not where this will be. Its drop dead naive to think that humans are better than computers in the long run at....anything.

Watson is already doing these type of things at Sloan kettering.

I will give the AI another 100 years and it won't be able to play Starcraft competitively lol. You have to understand the the program cannot be smarter than the programmers. And the smartest people don't do programming ^^

 

[libertyyne]

I'm pretty sure most their jobs have already been stolen by interventionists

and arent the interventionalists having to compete with plain jane cardiologists?

 

[LutGholein]

http://www.nature.com/articles/ncomms12474

Yet we already have research efforts to replace radiology and pathology...

It's simple
we kill the programmers.

 

[aldol16]

I hope the following provides some insight as to the automated versus human debate here:

There's a similar revolution going on in science - perhaps one that is more mature than the one in medicine. In organic chemistry, the utility of the organic chemist is hotly debated now - should chemists go into that field, since high-throughput machines have been created that can do essentially the work of 10 graduate organic chemistry students? I believe it was a Science paper that revealed this really amazing machine. Basically, the machine can weigh and mix in reagents, add the solvent, and quench the reaction whenever you want. And since it's fully automated, it can make precise measurements and so you can scale down your reaction so that you can do it in a 96-well plate and boom, you're in business. Now, you can run 96 reactions in a day using any permutation of reagents and solvent you want for a given reaction and screen for optimal conditions very quickly. (You'll be hard-pressed to find a graduate student who can set up and process 96 reactions a day).

So with all of this technology, what role does an organic chemist have? I believe that these new-fangled devices can be tools for the synthetic organic chemist instead of replacements. There are several reasons for this. Machines need guidance. You have to tell the machine what to look for and that takes expertise. If I want to do a C-H activation reaction, I have to start with reagents that are known to functionalize C-H bonds. I can't just mix in every reagent ever created. That takes insight and expertise. Machines also may take longer to find the solution. This is because machines can only do the laws of statistics and probability to do what they are programmed to do. As such, if you're looking for a path from A to C, the machine will necessarily apply a combinatorial approach to finding that path, testing various paths from A to C. But humans may have insightful shortcuts. You may already know from experience that passing through B makes the trip shorter. So you can tell the machine to compute only paths that go through B. Now you've drastically reduced the computational power needed by reducing the sample space.

Applied to medicine, this implies that physicians will be the benefactors of these new tools because they will enable them to perform at a much higher level. In contrast to replacing physicians, these machines will actually help physicians accomplish many more feats - scientifically and otherwise. Think of it as the PCR machine. Before the automated system, graduate students had to sit there and manually go through the PCR cycles. Think about the opportunity cost lost there. What else could they have been doing with that time devoted to menial labor? Physicians will always provide the expertise required to guide these machines and derive maximum benefit from them.

 

[7331poas]

I will give the AI another 100 years and it won't be able to play Starcraft competitively lol. You have to understand the the program cannot be smarter than the programmers. And the smartest people don't do programming ^^

You mean like IBM deep blue?

 

[Nommonnom]

You all have a conception of computers from about 5-10 years ago. Machines are able to get to orders of magnitude higher creative intelligence than people in a fraction of the time right now via machine learning. And yes that means the program can be smarter than the programmer.

 

[wizzed101]

You mean like IBM deep blue?

Chess is obviously the most complex game there is! What achievement! Human BFTO!!!

You see the people who are good at playing Startcraft don't work for IBM. They have better things to do. Chess on the other hand is pretty simple. But even so the way the computer played was pretty stupid because it reflected the mind of those who programmed it. If they are good at chess they wouldn't be working for IBM lmao. After you inject a bunch of grey matter (processing power) into a very stupid man, and his stupidity only becomes very efficient!

 

[7331poas]

Chess is obviously the most complex game there is! What achievement! Human BFTO!!!

You see the people who are good at playing Startcraft don't work for IBM. They have better things to do. Chess on the other hand is pretty simple. But even so the way the computer played was pretty stupid because it reflected the mind of those who programmed it. If they are good at chess they wouldn't be working for IBM lmao. After you inject a bunch of grey matter (processing power) into a very stupid man, and his stupidity only becomes very efficient!

If you think the computer is dumb, id hate to see what you think of Kasparov!

 

[wizzed101]

If you think the computer is dumb, id hate to see what you think of Kasparov!

He is very smart, too smart for the like of IBM. And there was allegation that they cheated. Not that the victory meant much. If it needs to test 200,000,000 moves per second to compete, it is grossly inefficient and is very stupid.

 

[Tenk]

My job stability as an EM physician is based on Anti-Darwinism and Murphy's Law.

We will never be obsolete.

Unless Skynet.

 

[Sardinia]

We may get to a point where machines are able to do everything, but they will still be unable to assume full culpability when it comes to an error.

 

[FootballFoot]

Even if a lot of machines are to help physicians work more efficiently, this will lower the demand for the specialty and make it a lot harder to find jobs since 1 doctor can do what 10 could in the past.

 

[FindersFee5]

I will give the AI another 100 years and it won't be able to play Starcraft competitively lol. You have to understand the the program cannot be smarter than the programmers. And the smartest people don't do programming ^^

Really? You don't think there are brilliant people working in programming? This is one of the more ignorant things I've ever heard.

Also, it doesn't matter HOW a computer wins, if a computer is better at making diagnoses than a human, it's better. It doesn't matter if it has a "stupid" method.

I hope the following provides some insight as to the automated versus human debate here:

There's a similar revolution going on in science - perhaps one that is more mature than the one in medicine. In organic chemistry, the utility of the organic chemist is hotly debated now - should chemists go into that field, since high-throughput machines have been created that can do essentially the work of 10 graduate organic chemistry students? I believe it was a Science paper that revealed this really amazing machine. Basically, the machine can weigh and mix in reagents, add the solvent, and quench the reaction whenever you want. And since it's fully automated, it can make precise measurements and so you can scale down your reaction so that you can do it in a 96-well plate and boom, you're in business. Now, you can run 96 reactions in a day using any permutation of reagents and solvent you want for a given reaction and screen for optimal conditions very quickly. (You'll be hard-pressed to find a graduate student who can set up and process 96 reactions a day).

So with all of this technology, what role does an organic chemist have? I believe that these new-fangled devices can be tools for the synthetic organic chemist instead of replacements. There are several reasons for this. Machines need guidance. You have to tell the machine what to look for and that takes expertise. If I want to do a C-H activation reaction, I have to start with reagents that are known to functionalize C-H bonds. I can't just mix in every reagent ever created. That takes insight and expertise. Machines also may take longer to find the solution. This is because machines can only do the laws of statistics and probability to do what they are programmed to do. As such, if you're looking for a path from A to C, the machine will necessarily apply a combinatorial approach to finding that path, testing various paths from A to C. But humans may have insightful shortcuts. You may already know from experience that passing through B makes the trip shorter. So you can tell the machine to compute only paths that go through B. Now you've drastically reduced the computational power needed by reducing the sample space.

Applied to medicine, this implies that physicians will be the benefactors of these new tools because they will enable them to perform at a much higher level. In contrast to replacing physicians, these machines will actually help physicians accomplish many more feats - scientifically and otherwise. Think of it as the PCR machine. Before the automated system, graduate students had to sit there and manually go through the PCR cycles. Think about the opportunity cost lost there. What else could they have been doing with that time devoted to menial labor? Physicians will always provide the expertise required to guide these machines and derive maximum benefit from them.

What you're describing is LITERALLY replacing jobs. Allowing somebody to do something more efficiently IS replacing jobs. Just because it isn't replacing every single person doesn't mean that people aren't losing jobs. If a pathologist and a computer could suddenly do the work of 10 pathologists, we don't need those extra 9. That's essentially reducing the demand for pathologists to 10% what it was. Are you really going to feel comfortable going into a field that could potentially not need 9/10 of its members in 30 years?

 

[FindersFee5]

Even if a lot of machines are to help physicians work more efficiently, this will lower the demand for the specialty and make it a lot harder to find jobs since 1 doctor can do what 10 could in the past.

whooops you beat me to my point and said it a lot more succinctly. Shoulda loaded new messages 🙁

 

[aldol16]

What you're describing is LITERALLY replacing jobs. Allowing somebody to do something more efficiently IS replacing jobs. Just because it isn't replacing every single person doesn't mean that people aren't losing jobs. If a pathologist and a computer could suddenly do the work of 10 pathologists, we don't need those extra 9. That's essentially reducing the demand for pathologists for 10%. Are you really going to feel comfortable going into a field that could potentially not need 9/10 of its members in 30 years?

I do not believe that replacing jobs is a necessary consequence of added efficiency. I think it allows roles to evolve and physicians to take on more responsibilities without making any of them obsolete. I use PCR as an example because most people on here are familiar with it in some degree. Do you feel that molecular biology graduate students have been replaced? Are PIs hiring fewer of them now because each graduate student can be more productive? If you're familiar with the field, that's a resounding no. In fact, there's a problem with having too many graduate students. Forty years ago, a graduate student had to spend a lot of time doing menial tasks. Now, that student can be more productive because of automation.

In short, arguing that added efficiency replaces jobs is only true if the magnitude of the task(s) performed by that job is small and invariable. If job A needs to get 50 tasks done in a day and the efficiency of employees is 5 tasks/day, then the employer will hire 10 employees. Your argument is that if efficiency increases to 10 tasks/day, then the employer necessarily only needs to hire 5 employees. 5 employees are now jobless. This is only true if that number - 50 tasks - doesn't change and is small. In fields like medicine and science, I would say that number isn't necessarily small - for medicine, there is a shortage at least in the primary care specialties. So if you need to get 150 tasks done in a day but now for some reason only have 10 employees working at a rate of 5 tasks/day, you're going to have a shortfall of 100 tasks that need doing. Now, the reason that you only have 10 employees may be that only 10 people have the qualifications for that job - the definition of a talent shortage. But now say that you invent a machine that can increase the efficiency of your employees to 10 tasks/day. Now, you can accomplish 100 tasks, leaving you with a shortfall of 50 tasks. Efficiency has improved, but you're not firing anyone because there are still tasks that need doing!

Now, the only thing that remains to be addressed is if efficiency improves to >15 tasks/day per employee. Then you could argue that the company can achieve 150 tasks per day and that's all it needs, so any high efficiency is a waste of labor. But this is assuming that the task dimensions don't change (i.e. that the task doesn't become more multi-faceted than previously imagined, with the advancement of science). In other words, your 150 tasks might transform into 300 tasks because science has advanced and the task is more complex than imagined before. I like when science is compared to trying to make a clearing in a dense forest. As you clear more and more trees, the perimeter of uncleared trees just increases. That is how I felt throughout my graduate studies. The more that we know, the more that we know we don't know. Is this the case for medicine? I don't know because I'm not an expert in medicine. Perhaps you could provide greater insight. But I can say that where medicine meets science, that analogy is very much true.

 

[Nommonnom]

.

 

[7331poas]

And i think alot of you dont understand the basic mechanisms of machine learning. the goal is that computers can gain knowledge the programmers do not instill directly. HOWEVER, even in current machine learning efforts with Watson, oncologists are regularly curating his content and decision making in the care plan process.

 

[Pepe18]

.

 

[sliceofbread136]

I honestly don't think anesthesiology is in trouble for a while. Yeah, a lot of it is reading graphs and pattern recognition, but there are also a lot more procedural aspects than many people think of (airway management, injections, post-procedural care, etc.)

It's much easier to teach a mid level a procedure than how to think through a complex problem

 

[libertyyne]

There is a large difference between strong AI and weak AI. What we currently have is weak AI and is

It's much easier to teach a mid level a procedure than how to think through a complex problem

A simple procedure yes. You don't see mid levels being taught how to do bypass,or appendectomy or even vertebraplasty.

 

[mvenus929]

This is an interesting question, and part of the reason I'm commenting is to follow this thread. But also, I've heard cardiothoracic surgery is at risk of becoming obsolete.

Nope. At least not on the congenital side. There are way too many procedures that can't be fixed by an interventionalist. And I've seen two caths go drastically wrong and need CT surgeon backup.

In the short term certainly, but by the end of our career, it is conceivable that there will be no new radiologists. My physiology professor told us stories about how a friend of his, during residency, would check all of the EKGs for accuracy. Now, that job doesn't seem to exist anymore. I could be wrong, but if it's true, I don't see why that wouldn't extend to radiology or anesthesiology.

All cardiologists check the machine for accuracy. For the run of the mill EKG, it doesn't usually matter, but they often find things that the machine doesn't see, or can dispute things that the machine does see. It's why we are all still taught how to read them.

 

[sullen-burger]

Considering how quickly medical technology has been moving in the past 16 years, should medical students be considering the possibility of their specialty being replaced, when picking a residency?

If so, which specialties are more susceptible to this form of obsolescence? Radiology and anesthesiology come to mind, but what about radiation oncology?

No. The economics of a work force is not a zero sum game. This literally is a fallacy called lump of labor.

Those specialties that won't adapt to survive selection.