Former FDA commissioner says AI may take on doctor roles sooner rather than later

[voxveritatisetlucis]

Op-ed: Gottlieb says A.I. may take on doctors' roles sooner rather than later

To grasp the potential of these tools to revolutionize medical care, it pays to start with a taxonomy of the different technologies and how they're being used.

www.cnbc.com

Thoughts

---

Members don't see this ad.

 

[Postictal Raiden]

Based on what he said, within 5 years (depending on regulations), we can see AI plays role in the simple encounters (the examples he gave MSK and URI stuff). The way I interpret this, midlevels are the first to be effected since they are occupying this role as of now.

If this truly happens, then the next phase would be replacing moderate to high level complexity type of encounters. This is where poop will hit the fan hard for us.

Thing though, if AI is going to drastically disrupt our careers, then what will happen to other more mundane jobs?Banking, pharmacy, finance, accounting, sales, data analysis, actuary, paralegal, etc? Will the world allow AI to replace two third, or more, of careers?

There will be very scary consequences when people lose their livelihood, their professional identity, and their sense of purpose.

 

[Mr.Smile12]

Watch the Hollywood WGA/SAG-AFTRA strikes (or join if you have friends)... that will be telling how we come to accept AI.

 

[bGMx]

The future is in general artificial intelligence. That said, in America, "security" is such a concern that many Americans will be hesitant to engage. There will likely be two populations of patients: those who rather see an AI, and those who don't. And for physicians: those who use AI and those who don't. I'm personally blown away by the number of medical students and residents who aren't aware of chatGPT.

 

[RangerBob]

We've been told we'd have full autonomous self-driving cars by now. Yet we still don't. Which seems far easier to me than teaching a computer to be someone's doctor.

Considering how slow the government is to regulate AI/tech, I also don't see the gov't acting to give AI a medical license anytime soon. It doesn't matter how good ChatCPT's board scores are if it's not licensed to practice medicine--until it has a license, at best all it can do is augment physicians/midlevels. And maybe act as a better Google/WebMD for patients trying to self-diagnose. Both of those do still have the potential to still disrupt the field if it makes us notably more efficient (I'm not so sure it would). I could see some physicians/midlevels losing jobs from that.

Still, as pointed out by Ibn Alfanis MD above, by then we'll have seen lots of society turmoil from all the more routine/mundane jobs that can be replaced prior to us (many/most of which wouldn't need a change in licensing/laws to be replaced). Lawyers and politicians are more vulnerable than physicians. So clearly AI will get addressed when the people who run the country's livelihoods are threatened. And I don't see the public voting to allow AI to govern us (even though if benevolent it could probably do a better job.)

AI is worrisome, but we're not the canaries in the mines, so things will either be sorted out by the time our field is dramatically affected, or things will have already really hit the fan. Look at how much society is already fraying now--imagine if even 1/10 the jobs out there get replaced by AI. What are all those people going to do? Go back to school for something else that will be replaced? All learn a trade at the same time?

 

[mistafab]

Literally zero chance of being replaced in the next 60 years.

 

[operaman]

Yeah definitely not worried. If anything, I’m excited to see how AI makes my job easier. Making the diagnosis is ridiculously easy 99% of the time. Most of my actual job is counseling and shared decision making with patients and families in addition to performing certain procedures. The back and forth aspect of helping someone decide how to proceed with their care really does require a human interface most of the time.

As for simple outpatient acute needs, I’m not sure we even really need AI for that. Much of the western world gives that job partly to pharmacists. I remember plenty of times working overseas and just walking into a pharmacy and talking to a pharmacist and leaving with antibiotics or cough medicine or a steroid dose pack or whatever. Even in the US we’ve given that job to nurses so it may easily be something an AI can handle in the near future. Though in reality, if the AI is mostly evidence based, it probably won’t hand out the abx and other meds so patients may not like it at all!

The key things where I want AI to help me:

1) reviewing and summarizing the whole record. I’m good at finding what I need but I’m sure I miss things. Would be nice to have a focused summary of all relevant information.

2) scribing notes. A good AI could even take an initial history and draft a note.

3) inbox management. My staff already do this, but it’s a burden and would be nice to have something that could respond 24/7 to easy questions and schedule follow up visits for anything requiring decision making.

 

[voxveritatisetlucis]

The future is in general artificial intelligence. That said, in America, "security" is such a concern that many Americans will be hesitant to engage. There will likely be two populations of patients: those who rather see an AI, and those who don't. And for physicians: those who use AI and those who don't. I'm personally blown away by the number of medical students and residents who aren't aware of chatGPT.

Yes but let’s say even 20% of patients were okay with seeing an AI, that would be significant in making it harder for everybody to get a job.

It’s like in radiology. I doubt that AI will replace radiologists but if 1 radiologist with AI can do the work of 5 without, then there will be extreme pressure on the job market

 

[VA Hopeful Dr]

Yes but let’s say even 20% of patients were okay with seeing an AI, that would be significant in making it harder for everybody to get a job.

It’s like in radiology. I doubt that AI will replace radiologists but if 1 radiologist with AI can do the work of 5 without, then there will be extreme pressure on the job market

Most fields will be fine. We're still not even half way through all the boomers retiring. In my specialty, over a third of practicing FPs are boomers. In fields that aren't expanding residency spots, I don't see finding jobs being an issue anytime soon.

 

[Approach]

From the video: What is this nonsense about AI showing more empathy than a human? Unfortunately, that wasn't even the most disappointing part of that video. Instead, it was the implication healthcare costs would get lowered by the approval of AI in healthcare.

For the foreseeable future, AI will augment the physician workflow, not replace the physician. Nonetheless, I won't lie. I am interested in choosing a specialty that is the "most AI resistant" simply because private equity will do anything to cut costs and probably won't hesitate to save money as soon as AI gets real traction.

Regarding regulatory pace, the callback to self-driving cars RangerBob is apt based on my experience. I worked for a significant automaker circa 2016, and all the talk was, "In 5-10 years, self-driving cars will be everywhere, putting cab drivers, truck drivers, etc. out of work and disrupting transit as we know it." Look at the self-driving car rhetoric now and the pace of their eventual adoption (it will be a while)...

All that said, if doctors are in trouble, I have yet to learn what my former colleagues when I worked in consulting would do for a job. I cannot tell you just how many hours I spent building decks, conducting Excel analysis, etc., when we're now at the point where one can skip the meeting and paste meeting notes into Bing AI and handle all the horsepower that previously a multitude of analysts, consultants, and senior consultants conducted.

Edit: If this is the study they are referring to, that's pretty misleading, IMO, as when I hear the word human (in this context, I pictured a psychiatrist), I picture face-to-face interaction, not how long a response is on Reddit.

 

[Tenk]

The first time an AI has to deal with a homeless meth addict it will format its own hard drive.

 

[pyrrion89]

the future of american AI healthcare:

(Elysium starring Matt Damon)

 

[NotAProgDirector]

The key things where I want AI to help me:

1) reviewing and summarizing the whole record. I’m good at finding what I need but I’m sure I miss things. Would be nice to have a focused summary of all relevant information.

2) scribing notes. A good AI could even take an initial history and draft a note.

3) inbox management. My staff already do this, but it’s a burden and would be nice to have something that could respond 24/7 to easy questions and schedule follow up visits for anything requiring decision making.

#2 is basically here now. Nuance DAX Express, Abridge, etc. They are not perfect, but they are pretty good.

#1 is coming. It may be slow because the dev is going to need to be the EMR companies. It's unlikely a 3rd party is going to be able to access everything in Epic, although I guess with enough FHIR feeds it's possible.

#3 is also underway by Epic already. Very basic so far, but they are just starting to leverage LLM's so we will see.

 

[operaman]

#2 is basically here now. Nuance DAX Express, Abridge, etc. They are not perfect, but they are pretty good.

#1 is coming. It may be slow because the dev is going to need to be the EMR companies. It's unlikely a 3rd party is going to be able to access everything in Epic, although I guess with enough FHIR feeds it's possible.

#3 is also underway by Epic already. Very basic so far, but they are just starting to leverage LLM's so we will see.

Dang how did I not know about some of these! The ambient AI helping with documenting will be huge, and exactly the kind of physician task that would be well suited to AI and LLMs.

Would definitely allow me to see a lot more patients each day. As it is I purposely end earlier so I have enough time to document and leave at a reasonable time. I’ll have to see if I can convince our IT folks to invest in something like this.

 

[NotAProgDirector]

Dang how did I not know about some of these! The ambient AI helping with documenting will be huge, and exactly the kind of physician task that would be well suited to AI and LLMs.

Would definitely allow me to see a lot more patients each day. As it is I purposely end earlier so I have enough time to document and leave at a reasonable time. I’ll have to see if I can convince our IT folks to invest in something like this.

We're looking at this now. Probably only need to see 1-2 more patients per day to pay for it.

 

[ChordaEpiphany]

We've been told we'd have full autonomous self-driving cars by now. Yet we still don't. Which seems far easier to me than teaching a computer to be someone's doctor.

Considering how slow the government is to regulate AI/tech, I also don't see the gov't acting to give AI a medical license anytime soon. It doesn't matter how good ChatCPT's board scores are if it's not licensed to practice medicine--until it has a license, at best all it can do is augment physicians/midlevels. And maybe act as a better Google/WebMD for patients trying to self-diagnose. Both of those do still have the potential to still disrupt the field if it makes us notably more efficient (I'm not so sure it would). I could see some physicians/midlevels losing jobs from that.

Still, as pointed out by Ibn Alfanis MD above, by then we'll have seen lots of society turmoil from all the more routine/mundane jobs that can be replaced prior to us (many/most of which wouldn't need a change in licensing/laws to be replaced). Lawyers and politicians are more vulnerable than physicians. So clearly AI will get addressed when the people who run the country's livelihoods are threatened. And I don't see the public voting to allow AI to govern us (even though if benevolent it could probably do a better job.)

AI is worrisome, but we're not the canaries in the mines, so things will either be sorted out by the time our field is dramatically affected, or things will have already really hit the fan. Look at how much society is already fraying now--imagine if even 1/10 the jobs out there get replaced by AI. What are all those people going to do? Go back to school for something else that will be replaced? All learn a trade at the same time?

I think people are being a little too relaxed about AI. As a pure clinical physician, especially in non-procedural specialties, your job is much harder to do by another human than some random admin job. However, it's also way more standardized. Standardized work performed by a large group with high salaries is a dream for AI companies.

Someone who works in a cubicle at a big bank is doing boring, easy work all day long, but it's very different every day. It's "email Brian to get the reports from Judy so that you can reformat them the way Boss Man likes it so he can present to Bigger Boss Man." The next day it's something different, like "analyze this client's portfolio, sent to you in some random mix of PDFs and excel sheets with no standardized format, to determine if they should get a loan." You could pull any schmuck with a business degree off the street and teach them the job in a week. However, it's simply not worth it for a company to automate all that without general AI. It's too logistically cumbersome for a low payout, because most of these workers spend most of their time doing simple but unique tasks. Directing the AI would be as work intensive as just hiring someone to do the work.

On the other hand, a community physician is doing really similar stuff in every office across the US. It's why locums is a thing. Give an AI a good H&P and it will generate something eerily similar to an attending's assessment and plan. It's not going to replace the physician, but it sure might close the gap between physician and midlevel. Imo physicians are being foolish by insisting on being the hands-on care provider rather than a people manager with technical expertise. From a purely monetary/relevance perspective, the right approach is to reorient physicians as midlevel managers who take only high acuity cases. Then cut the supply drastically, anticipating the flood of midlevels and the inevitable change to the care model. Very few top engineers are still CAD'ing in their 50s. Eventually they move into management roles where their technical expertise is still valuable and necessary, but it isn't their day-to-day. Physicians are the only technical experts that continue to handle the bread and butter until retirement.

However, I know the above is deeply unpopular with physicians. Every time I suggest it online I get absolutely reamed out. Physicians like being "the doctor." However, we are inevitably entering a world where performing repetitive intellectual labor will no longer be valuable. It's really time to start thinking about maximizing unique tasks. Part of that will be leaving behind the bread and butter, embracing a role as a manager, and lobbying for appropriate compensation for tasks that are not achievable with AI.

 

[voxveritatisetlucis]

I think people are being a little too relaxed about AI. As a pure clinical physician, especially in non-procedural specialties, your job is much harder to do by another human than some random admin job. However, it's also way more standardized. Standardized work performed by a large group with high salaries is a dream for AI companies.

Someone who works in a cubicle at a big bank is doing boring, easy work all day long, but it's very different every day. It's "email Brian to get the reports from Judy so that you can reformat them the way Boss Man likes it so he can present to Bigger Boss Man." The next day it's something different, like "analyze this client's portfolio, sent to you in some random mix of PDFs and excel sheets with no standardized format, to determine if they should get a loan." You could pull any schmuck with a business degree off the street and teach them the job in a week. However, it's simply not worth it for a company to automate all that without general AI. It's too logistically cumbersome for a low payout, because most of these workers spend most of their time doing simple but unique tasks. Directing the AI would be as work intensive as just hiring someone to do the work.

On the other hand, a community physician is doing really similar stuff in every office across the US. It's why locums is a thing. Give an AI a good H&P and it will generate something eerily similar to an attending's assessment and plan. It's not going to replace the physician, but it sure might close the gap between physician and midlevel. Imo physicians are being foolish by insisting on being the hands-on care provider rather than a people manager with technical expertise. From a purely monetary/relevance perspective, the right approach is to reorient physicians as midlevel managers who take only high acuity cases. Then cut the supply drastically, anticipating the flood of midlevels and the inevitable change to the care model. Very few top engineers are still CAD'ing in their 50s. Eventually they move into management roles where their technical expertise is still valuable and necessary, but it isn't their day-to-day. Physicians are the only technical experts that continue to handle the bread and butter until retirement.

However, I know the above is deeply unpopular with physicians. Every time I suggest it online I get absolutely reamed out. Physicians like being "the doctor." However, we are inevitably entering a world where performing repetitive intellectual labor will no longer be valuable. It's really time to start thinking about maximizing unique tasks. Part of that will be leaving behind the bread and butter, embracing a role as a manager, and lobbying for appropriate compensation for tasks that are not achievable with AI.

But it is unlikely that all physicians would be able to get one of these managerial type jobs.

I wonder if surgery and surgical subs will get much more competitive

 

[ChordaEpiphany]

But it is unlikely that all physicians would be able to get one of these managerial type jobs.

I wonder if surgery and surgical subs will get much more competitive

This is why we should dramatically cut residencies and lean into midlevels filling the gap. Physicians could act as high acuity case managers and low acuity people managers. The reason that midlevels are successfully moving into the physician space is that mediocre people can do a fine job billing for a visit. Attending MDs have the attributes of other high-level technical managers like the upper levels of the law, consulting, finance, and corporate world. Those positions are not gate-keeped by board certifications, yet demand for top talent keeps compensation very high. Similarly, midlevels could never compete with an MD for the medicine-equivalent role.

As for competitiveness, I think it will be driven by incredibly short-sighted views. If rads goes WFH and imaging throughput goes through the roof, with comp rising, people will flood in even if the 30 year horizon looks awful. Same for many intellectual specialties. For a short time this will likely improve the QOL of non-proceduralists. Surgical subs will remain competitive because they already are, but I think it will take a report like the one for EM to dissuade people from non-procedural specialties.

 

[deleted1139416]

This is why we should dramatically cut residencies and lean into midlevels filling the gap. Physicians could act as high acuity case managers and low acuity people managers. The reason that midlevels are successfully moving into the physician space is that mediocre people can do a fine job billing for a visit. Attending MDs have the attributes of other high-level technical managers like the upper levels of the law, consulting, finance, and corporate world. Those positions are not gate-keeped by board certifications, yet demand for top talent keeps compensation very high. Similarly, midlevels could never compete with an MD for the medicine-equivalent role.

As for competitiveness, I think it will be driven by incredibly short-sighted views. If rads goes WFH and imaging throughput goes through the roof, with comp rising, people will flood in even if the 30 year horizon looks awful. Same for many intellectual specialties. For a short time this will likely improve the QOL of non-proceduralists. Surgical subs will remain competitive because they already are, but I think it will take a report like the one for EM to dissuade people from non-procedural specialties.

I think we need to look for the positives in AI and how they will improve healthcare. Anyone can look at a cholesterol level and prescribe a statin. But you and I know that health is much more complicated than that. I am hoping that by getting rid of basic, routine stuff and all the labor that goes into charting, AI will free up doctor's time to actually focus on improving quality of life. The current state of healthcare in the united states is very very very bad for many minority communities. AI is going to make that better.

 

[qwerty89]

Employed physicians will get screwed 100%. Any additional profit from efficiency gains will go to the hospital, CMG, private equity group, etc.

For rads, those in true private practices where they collect on each case will probably have one last golden age where they will reap the benefits of increased productivity due to AI (if they embrace change early) before reimbursement per case falls through the floor.

 

[ChordaEpiphany]

I think we need to look for the positives in AI and how they will improve healthcare. Anyone can look at a cholesterol level and prescribe a statin. But you and I know that health is much more complicated than that. I am hoping that by getting rid of basic, routine stuff and all the labor that goes into charting, AI will free up doctor's time to actually focus on improving quality of life. The current state of healthcare in the united states is very very very bad for many minority communities. AI is going to make that better.

I hate to be morbid about it, but doctor's lack of free time and charting isn't the reason healthcare is so abysmal. Charting is actually just a symptom of the disease. If doctors get more free time, it won't be spent counseling the underserved and forming therapeutic alliances. Admins will push for more productivity and CMS will suppress payments, forcing doctors to see more patients/hour.

Really we should have done away with basic/routine stuff for doctors long ago. Like I said above, we should have leaned into mid-levels as bread and butter providers, with a doctor taking the higher acuity patients (and always present for back up when necessary). No one would be talking about doctor's being on the chopping block if we had structured medicine like that. They'd be looking to replace mid-levels and make doctors AI-managers. However, the pseudo-planned economy of healthcare created tons of perverse incentives that resulted in doctors fighting with mid-levels for bread and butter instead of dividing and conquering in a logical way. CMS (and thus also private insurance) pays more for certain complex issues, but not enough to offset the time required.

I think the potential is there for AI to improve healthcare for patients, but ultimately admins will push so hard on doctors that it'll turn into a spitting match over pay. Doctors will be too wrapped up in their own problems losing ground as respected leaders in healthcare to properly advocate for patients, and the few that do will be offered a severe pay cut by their MBA bosses. It's going to be all about efficiency, and patients will feel more than ever like cattle being herded through the system, screened with an iPad and sent on their way by a disgruntled, overworked nurse.

 

[greatnt249]

Like I said above, we should have leaned into mid-levels as bread and butter providers, with a doctor taking the higher acuity patients (and always present for back up when necessary).

1. There's nothing "bread and butter" about medicine; it might seem "easy" or "routine" to the outside observer, but it only becomes that way through experience and training. Our training is for when what outwardly appears to be easy is not, in fact, easy or straightforward.

2. Seeing only high-acuity cases all the time is a quick recipe for burnout.

 

[deleted1139416]

Cope harder.

ChatGPT Out-scores Medical Students on Complex Clinical Care Exam Questions

A new study shows AI's capabilities at analyzing medical text and offering diagnoses — and forces a rethink of medical education.

hai.stanford.edu

Did you even read what you linked? It compared it to 1st and 2nd year med students. Everyone knows that 1st and 2nd year med students don’t know jack. Even 3rd/4th year students don’t know jack.

When ChatGPT took the USMLE, it barely scored above the cutoff for a pass, just above 60%. Considering that nearly 95% of students pass USMLE on the first try, your claims are just fear mongering and nothing else.

 

[JamalJ]

Did you even read what you linked? It compared it to 1st and 2nd year med students. Everyone knows that 1st and 2nd year med students don’t know jack. Even 3rd/4th year students don’t know jack.

When ChatGPT took the USMLE, it barely scored above the cutoff for a pass, just above 60%. Considering that nearly 95% of students pass USMLE on the first try, your claims are just fear mongering and nothing else.

But ChatGPT is just a few years old. The field of AI is progressing exponentially and ChatGPT 4 is a lot better than the 3rd version (I have tried both; they are just a few months apart). So in a few years, things are going to change dramatically.

Personally, I feel it is very likely that physicians will become like pilots. They basically oversee everything, explain to the passangers, etc. I guess we will still have to do physical exams, but even that can be replaced. Our main job will probably be to choose therapeutic options, make things simple for the patient, etc.

Personally, I think it will become a lot less desirable job, more menial probably. Sort of like a nurse, I guess. Salaries will go down. The status of a physician has been in decline and things will get worse, imo.

You are coping real hard, that is for sure.

 

[JamalJ]

Seeing how common sense is rarely common these days, I was curious as to what experiences led you to your opinion(s).

I did research with a surgeon over the summer and saw how he did surgeries. A lot of it is working with nurses, anesthesia, techs, etc. Lots of physical work, so that's hard for a robot to replace.

Of course, I may be coping myself. It may well be the case that surgeons get replaced too. I can easily see how radiology and anesthesiology will be the first to go, as they are very technical and use little judgment. Pathology too. They are done, imo.

 

[greatnt249]

I did research with a surgeon over the summer and saw how he did surgeries. A lot of it is working with nurses, anesthesia, techs, etc. Lots of physical work, so that's hard for a robot to replace.

Of course, I may be coping myself. It may well be the case that surgeons get replaced too. I can easily see how radiology and anesthesiology will be the first to go, as they are very technical and use little judgment. Pathology too. They are done, imo.

So, you did research with a surgeon over a summer and can now project what the future of AI is in medicine? Why so angry with my responses?

 

[RangerBob]

I did research with a surgeon over the summer and saw how he did surgeries. A lot of it is working with nurses, anesthesia, techs, etc. Lots of physical work, so that's hard for a robot to replace.

Of course, I may be coping myself. It may well be the case that surgeons get replaced too. I can easily see how radiology and anesthesiology will be the first to go, as they are very technical and use little judgment. Pathology too. They are done, imo.

The day our lives are literally in the hands of a computer instead of an anesthesiologist, you will have no patients ever getting surgery again (unless it's emergent and consent isn't required). Not happening until we trust AI to be our teachers, leaders, accountants, drivers, pilots, and spiritual leaders.

Cars will drive all of us around well before that. And we are still a long ways off from cars driving us around safely, and even longer before mass adoption (particularly in more conservative areas). I'm not convinced it'll ever happen unless mandated by the government, and well... this is America, so that's not happening.

The same goes for radiology and pathology. AI can't legally practice medicine until it has a license and malpractice insurance. No state medical board is going to line up anytime soon to grant computers a license to practice medicine. And no state legislature will pass laws allowing it anytime soon (legislatures tend to lag behind the times with tech in the first place).

So most jobs are safe for some time. The more realistic threat is AI augmenting the work of physicians and either eliminating positions because one radiologist does the work of three now, or because we get asked to do more and more/intense work (like seeing only the really sick patients) and burn out.

As I've said before--politicians and lawyers will be replaced far before we do. And because they have the most power in the country, the issues with AI will be addressed by the time their jobs are threatened. That, or SkyNet has arrived and we have bigger things to worry about.

 

[Postictal Raiden]

The day our lives are literally in the hands of a computer instead of an anesthesiologist, you will have no patients ever getting surgery again (unless it's emergent and consent isn't required). Not happening until we trust AI to be our teachers, leaders, accountants, drivers, pilots, and spiritual leaders.

Cars will drive all of us around well before that. And we are still a long ways off from cars driving us around safely, and even longer before mass adoption (particularly in more conservative areas). I'm not convinced it'll ever happen unless mandated by the government, and well... this is America, so that's not happening.

The same goes for radiology and pathology. AI can't legally practice medicine until it has a license and malpractice insurance. No state medical board is going to line up anytime soon to grant computers a license to practice medicine. And no state legislature will pass laws allowing it anytime soon (legislatures tend to lag behind the times with tech in the first place).

So most jobs are safe for some time. The more realistic threat is AI augmenting the work of physicians and either eliminating positions because one radiologist does the work of three now, or because we get asked to do more and more/intense work (like seeing only the really sick patients) and burn out.

As I've said before--politicians and lawyers will be replaced far before we do. And because they have the most power in the country, the issues with AI will be addressed by the time their jobs are threatened. That, or SkyNet has arrived and we have bigger things to worry about.

Although radiology appears to be the low hanging fruit for AI takeover/augmentation, history tells that with every new innovation, the field has become even more essential than ever. More diagnostic technologies simply means clinicians ordering more pointless tests.

An example from my field (neurology), patient comes in for syncope. They get a CTH (that’s everyone’s admission ticket nowadays), then CTA, then an MRA, then a carotid US. Arguably, you didn’t need any of these tests to help you with the etiology. At most, I would get a CTA or an MRA. The carotid US is useless because you need to evaluate the vertebrobasilar system, not the carotid.

If a new diagnostic technology that can better diagnose syncope, I assure you we will still be ordering CTAs, MRAs and carotid USs.

 

[JamalJ]

So, you did research with a surgeon over a summer and can now project what the future of AI is in medicine? Why so angry with my responses?

Lol. Obviously, I can't predict but that is sort of what I think. But yeah I'd imagine it is hard to develop a robot that does surgery. Clinical I can see, but surgery, at least for me, is hard to see.

 

[JamalJ]

Although radiology appears to be the low hanging fruit for AI takeover/augmentation, history tells that with every new innovation, the field has become even more essential than ever. More diagnostic technologies simply means clinicians ordering more pointless tests.

An example from my field (neurology), patient comes in for syncope. They get a CTH (that’s everyone’s admission ticket nowadays), then CTA, then an MRA, then a carotid US. Arguably, you didn’t need any of these tests to help you with the etiology. At most, I would get a CTA or an MRA. The carotid US is useless because you need to evaluate the vertebrobasilar system, not the carotid.

If a new diagnostic technology that can better diagnose syncope, I assure you we will still be ordering CTAs, MRAs and carotid USs.

Why are they ordered?

 

[Lawpy]

I don’t understand why we can’t delegate charting and clerical duties to AI with some modifications and let doctors focus solely on clinical duties.

 

[GoSpursGo]

A number of posts from a troll who has been banned have been deleted, along with posts that were quoting those posts. I left a few posts from the troll to avoid the thread becoming unreadable.

 

[Postictal Raiden]

A number of posts from a troll who has been banned have been deleted, along with posts that were quoting those posts. I left a few posts from the troll to avoid the thread becoming unreadable.

Sorry couldn’t help it

 

[mmchick]

What I would really love /s with this technology is to use it on certain types of patients to gather the full H&P and whittle it down to a couple sentences. I'm talking unleashing this thing on the whole world is out to get me patient, no one in a 1000 mile radius has an answer patient, you can just look that up in my file patient, I lied about my health so I could get a pan-scan patient. Now THAT would be somethin.

 

[RangerBob]

What I would really love /s with this technology is to use it on certain types of patients to gather the full H&P and whittle it down to a couple sentences. I'm talking unleashing this thing on the whole world is out to get me patient, no one in a 1000 mile radius has an answer patient, you can just look that up in my file patient, I lied about my health so I could get a pan-scan patient. Now THAT would be somethin.

That's called a med student

Ok, actually that would be if you want things whittled down to a few paragraphs.

What you need is a senior resident

One serious note, I would worry about what AI would leave out. AI is already known for making stuff up. I certainly wouldn't trust AI to decide what information is relevant or not. At least I wouldn't trust it anytime soon until it proves it can drive, recognize the difference between a real apple and a billboard apple, etc.

 

[deleted1168269]

Although radiology appears to be the low hanging fruit for AI takeover/augmentation, history tells that with every new innovation, the field has become even more essential than ever. More diagnostic technologies simply means clinicians ordering more pointless tests.

An example from my field (neurology), patient comes in for syncope. They get a CTH (that’s everyone’s admission ticket nowadays), then CTA, then an MRA, then a carotid US. Arguably, you didn’t need any of these tests to help you with the etiology. At most, I would get a CTA or an MRA. The carotid US is useless because you need to evaluate the vertebrobasilar system, not the carotid.

If a new diagnostic technology that can better diagnose syncope, I assure you we will still be ordering CTAs, MRAs and carotid USs.

Wouldn’t the MRA give you info on carotid stenosis anyways? Why carotid US?