AI and specialty selection

[deleted746658]

i am curious to know what effect AI and its discourse is having on specialty selection, especially students early in their medical education

i will be entering radiology. the hype around AI/ML almost made me opt for a different field despite knowing well enough that DR best suited my interests.

for pre-clinical students/ undecided MS3/MS4s, how much will you let potentially disruptive technologies influence your selection of specialty?

---

Members don't see this ad.

 

[ArdorAyurveda]

I am interested in psychiatry, so I'm not really worried about AI

Interesting thread though, I was having this same thought last week

I highly doubt AI will replace physician labor, so I don't think it's something to be concerned about. Getting sound medical advice is a little different than truck driving, working in an automobile manufacturing plant, or working in the retail business.

 

[getdown]

i am curious to know what effect AI and its discourse is having on specialty selection, especially students early in their medical education

i will be entering radiology. the hype around AI/ML almost made me opt for a different field despite knowing well enough that DR best suited my interests.

for pre-clinical students/ undecided MS3/MS4s, how much will you let potentially disruptive technologies influence your selection of specialty?

Key phrase in your entire post is "potential". Medicine is changing all the time and new interventions and other specialty encroachment will siphon patients from your specialty. Making a decision based on fear of something happening in the future is ABSOLUTELY the wrong way of deciding. You need to base it on if you enjoy doing a particular specialty and can see yourself doing it for a long time. If I had let these thoughts cross my mind when I was choosing a specialty I would have steered clear of Anesthesia because of CRNA encroachment and all that ****. People have been bitching about CRNAs since the 70s and Anesthesia is still around. You still make good money and I'm still very happy doing it. Honestly can't really see myself doing anything else. You seriously can't predict what's going to happen in the future so just choose what's right for you now. Specialties adapt, and so will you. For what it's worth AI technology that's dependable and can diagnose things 100% of the time is so far away that it'll probably won't happen, if at all, till after you're close to being done with residency.

 

[kb1900]

I am considering rads and derm BECAUSE of the advances that AI can make in the fields.


These days computer programming, python, tensorflow, keras, Opencv etc. are so ridiculously accessible that I think it’s naive that to say “things are so far off”. But it’s equally ridiculous to assume the bread and butter of radiology will change significantly or “AI will take away jobs”. There’s plenty of examples in current fields where preliminary AI imaging screening or database-adaptive models are already in place. In the context of healthcare, AI is in its infancy and the future is exciting, not worrying.

 

[puahate]

Go rads then IR.

 

[7331poas]

Key phrase in your entire post is "potential". Medicine is changing all the time and new interventions and other specialty encroachment will siphon patients from your specialty. Making a decision based on fear of something happening in the future is ABSOLUTELY the wrong way of deciding. You need to base it on if you enjoy doing a particular specialty and can see yourself doing it for a long time. If I had let these thoughts cross my mind when I was choosing a specialty I would have steered clear of Anesthesia because of CRNA encroachment and all that ****. People have been bitching about CRNAs since the 70s and Anesthesia is still around. You still make good money and I'm still very happy doing it. Honestly can't really see myself doing anything else. You seriously can't predict what's going to happen in the future so just choose what's right for you now. Specialties adapt, and so will you. For what it's worth AI technology that's dependable and can diagnose things 100% of the time is so far away that it'll probably won't happen, if at all, till after you're close to being done with residency.

anesthesia may still be around, but CRNAs are a very real problem that has arisen since the 70s. Have CRNAs increased the wealth of anesthesia? Sure. Has CRNAs hurt the value of an anesthesiologist who actually likes doing anesthesia? Maybe a little.

You absolutely should consider the future of the specialty before deciding. I personally will not consider pathology because of the job market. I love pathology because of pathoma. I think it could be a cool field. But at the end of the day I will not even rotate in the field because I know what the market looks like.

 

[ciestar]

Personally, I am more worried about midlevels.

 

[Matthew9Thirtyfive]

There’s a good thread on the rads forum where a poster explains why AI is not a threat. It seemed well reasoned.

Here

 

[getdown]

anesthesia may still be around, but CRNAs are a very real problem that has arisen since the 70s. Have CRNAs increased the wealth of anesthesia? Sure. Has CRNAs hurt the value of an anesthesiologist who actually likes doing anesthesia? Maybe a little.

You absolutely should consider the future of the specialty before deciding. I personally will not consider pathology because of the job market. I love pathology because of pathoma. I think it could be a cool field. But at the end of the day I will not even rotate in the field because I know what the market looks like.

I understand your point but my point was that if you can't see yourself doing anything else besides that field, does it really make sense to go into another field just because it may be more "stable"? If someone is between 2 fields that they're really interested in one safe from midlevel encroachment and one that is rife with it than the smart choice would be the former. But if you can't see doing anything else? What then? Go into a field you'd be miserable in instead?

Midlevels are encroaching in all fields except maybe surgical ones. That literally is half of all medical fields and specialties. You would do well to at least consider it and take it into account. Obviously going in blind is stupid but if you've weighed your options and still found that despite this "threat" you'd still deal with that than taking 36 hour calls and surgeries? Or deal with dispo issues and crap ton of paperwork in IM? Yeah, I'd rather deal with CRNAs than any of that other stuff.

 

[Man o War]

I understand your point but my point was that if you can't see yourself doing anything else besides that field, does it really make sense to go into another field just because it may be more "stable"? If someone is between 2 fields that they're really interested in one safe from midlevel encroachment and one that is rife with it than the smart choice would be the former. But if you can't see doing anything else? What then? Go into a field you'd be miserable in instead?

Midlevels are encroaching in all fields except maybe surgical ones. That literally is half of all medical fields and specialties. You would do well to at least consider it and take it into account. Obviously going in blind is stupid but if you've weighed your options and still found that despite this "threat" you'd still deal with that than taking 36 hour calls and surgeries? Or deal with dispo issues and crap ton of paperwork in IM? Yeah, I'd rather deal with CRNAs than any of that other stuff.

Yep.
And single payer/rationing of care will cause surgery to take a hit. It’s coming, just a matter of when.
Every field in medicine has very real threats- so pick your poison.

 

[Man o War]

No way this country gets single payer. Too many people with say would lose too much money. No chance

Sure we will. We will have one universal payer eventually with an option to get private if you can afford it. Everyone gets insurance and Private Insurance companies can cherry pick like they used to before ACA. Everyone who “matters” is happy.
When that happens, you will have rationing of joint replacements like they do in the UK, and waits for testing and elective surgeries like they do in Canada. Honestly we should’ve started aggressively rationing care a long time ago, but that’s another discussion.

 

[Rekt]

Medicine as we know it will long die before AI takes over because mid-levels and our idiot colleague medical students calling for Medicare for all.

 

[differentiating]

I know a decent number of rads attendings who feel pretty confident that AI won’t replace them - in neurorads, they’re hoping AI will get good enough to eval for GK so they don’t have to waste the time, but it’s not there yet. Most see it as a potential tool in their toolbox and not a real threat. Which is fair - I can’t imagine most doctors (or hospitals) being willing to make treatment decisions without having an actual person there to confirm the read.

For me personally, I didn’t think about it at all. But it’s not as common a concern in my chosen field in general.

 

[operaman]

Yeah I used to think AI would replace rads but have come to change that viewpoint. It will almost certainly change the field and provide an incredible tool for radiologists to use. When I look at a radiology read it’s after I’ve already read the scan myself and I’m looking to make sure I didn’t miss anything else important since I’m usually looking with a singular focus and to confirm that my own impression is sound. When I call or go to the reading room or present scans at tumor board it’s because I need to have a high level conversation with another physician who’s better than me at reading scans, who can discuss and understand my clinical dilemma and offer an opinion or recommend additional imaging that would help answer the question.

I think AI will certainly help radiologists manage their workflow better and provide reassurance and error checking. It will hopefully help them turn around reads faster and spend more of their time focused on the truly meaningful and nuanced things.

 

[Cognovi]

I know a decent number of rads attendings who feel pretty confident that AI won’t replace them - in neurorads, they’re hoping AI will get good enough to eval for GK so they don’t have to waste the time, but it’s not there yet. Most see it as a potential tool in their toolbox and not a real threat. Which is fair - I can’t imagine most doctors (or hospitals) being willing to make treatment decisions without having an actual person there to confirm the read.

For me personally, I didn’t think about it at all. But it’s not as common a concern in my chosen field in general.

What is eval for GK

 

[sumtimesuwonder]

I don't think you should be worried about AI entirely replacing the field. I would worry about AI making a radiologists job easier and making it so that less radiologists are needed. Someone is going to have to operate the AI/ double check it no matter what. Its just how much quicker/more efficient will radiologists get with the advent of new technology and how many jobs will be taken away as a result? It could be none for all I know, maybe imaging modalities gets safer/more effective as a result of technology advances and more studies are performed, meaning more radiologists. I could see it going either way.

I'm not saying any of this will happen anytime soon, or even during any of our careers. But I am saying that it could happen eventually.

 

[differentiating]

What is eval for GK

Evaluate for gamma knife - aka just identifying dots of potential mets and labeling them all. All the neuroradiologists hated it.

 

[xALIAx]

asking this question on SDN is dumb because 95% of SDN posters repeat the "not in my lifetime" mantra with their heads stuck in the sand

look up 2-tiered healthcare and watch Idiocracy if you are interested

 

[ArdorAyurveda]

asking this question on SDN is dumb because 95% of SDN posters repeat the "not in my lifetime" mantra with their heads stuck in the sand

look up 2-tiered healthcare and watch Idiocracy if you are interested

What is the mechanism by which you expect artificial intelligence to negatively impact physician reimbursement or wages?

Idiocracy is a fictional movie and 2-tiered healthcare just means that poor people will get rationed healthcare while rich people get whatever they want (not much different from how it is now...)

What does these things have to do with artificial intelligence? I suppose there's the current fixation on the fiction of AI RoboDoctors?

I'm not worried.

That being said, I do think it would be wise for physicians to have at least a basic understanding of machine learning, learning how to code algorithms, understanding cybersecurity, reading about population health, etc. Physicians should learn these things because we will need to navigate an increasingly technological healthcare industry. It has nothing to do with the fear of losing my job to AI.

 

[Cognovi]

What is the mechanism by which you expect artificial intelligence to negatively impact physician reimbursement or wages?

Some computer program comes out that makes pulmonary nodule detection and reporting 50% faster. Some practices use it and are able to read a chest CT in a shorter amount of time and make more money. Eventually, most practices start using the program because it's so effective and efficient. Then, when the specialty society and the relative value unit update committee surveys the average practicing radiologist how much time it takes them to read a chest CT, that number is found to be lower than before. They also notice that the cost of the program is lower than the amount of extra money people make when they use the program. Then the committee recommends that the number of work relative value units assigned to a chest CT be reduced. This plays out over many years, because the diffusion of innovation is slow. In this time, radiologists make more money than before, and then less money. It evens out eventually to some degree. What people are not sure about is how rapid this shift is, whether on balance radiologists come out on top, whether the efficiency gains will be sufficiently balanced by an increase in demand for imaging in order to employ the same number of radiologists, and whether the field is able to modify in advance the number of residency positions available to accomodate changes in workforce demands.