DNPs will eventually have unlimited SOP

[grt398]

Why would they be forbidden? You could track things like consultations if you wanted (and compare those as well, since they factor into the cost).

In any case, if it is indeed impossible, then the games over. Can't be won. So someone better figure out a way to make it possible. Whether you agree with it or not, no one is going to (least of all elected officials - the ones who've given physicians a monopoly on certain types of procedures and scope of practice) just trust your word when the other side is coming up with data and potential solutions to the ever tightening budget problems.

So there's a few reasons I can't imagine them occurring:

1. The legal department of any large system would never allow such a thing to happen. The potential loss via lawsuits of allowing people to potentially die (let's say we use ICU patients) b/c they were randomized to a segment of a trial that forbids physician involvement would be enormous.

2. So let's say we include a stipulation that any consultation of physicians constitutes a "treatment failure" on the part of the DNP instead to get around #1 above. Where do you draw the line? Do you actually wait for a patient to code? What about a patient coming in for a dermatologic problem who gets misdiagnosed and shows up 6 months later with metastatic stage IV melanoma that could have been treated if it was caught at the initial visit?

3. Inability to get a broad spectrum of patients - let's be a bit practical; the odds of anyone with insurance agreeing to be randomized to care in a situation where they are told there will be no physician oversight unless they are about to die is going to be slim. Most patients will end up being poor and uninsured. Not good for broad applicability, and from a PR standpoint, a massive problem.

Those are just the first few things off the top of my head. I'm sure there are many other issues not being addressed. In the end I suspect it'll be lawsuits against DNPs that brings the issue under control, but a lot of patients are going to be hurt along the way.

---

Members don't see this ad.

 

[evilbooyaa]

But that's the thing, we're fine with PA and DNPs where they are, working with and under MD supervision.

Really the only thing that should be being discussed here is the DNPs working on their own without MD supervision. I'm ok with Midlevels midleveling.

 

[xenotype]

But that's the thing, we're fine with PA and DNPs where they are, working with and under MD supervision.

Really the only thing that should be being discussed here is the DNPs working on their own without MD supervision. I'm ok with Midlevels midleveling.

I don't think anyone has a problem with this. The midlevels are not happy to remain midleveling. They want more money and autonomy, and they will use any route to get it except going to medical school. PAs want it just as bad, but face much bigger political barriers.

 

[evilbooyaa]

I don't think anyone has a problem with this. The midlevels are not happy to remain midleveling. They want more money and autonomy, and they will use any route to get it except going to medical school. PAs want it just as bad, but face much bigger political barriers.

Sorry, that was more of a response to peteB.

agree, physicians aren't going to be united anytime soon. They are always fighting amongst themselves.

If we really wanted to fight this all physicians could stand together and say we are not going to hire or work with any PAs or DNPs. But that's never going to happen. There will always be some physicians who see that as an opportunity to make more money and ruin the entire movement.

I don't care about working WITH PAs or DNPs. If we say we aren't going to work with them, they'll start their own practices. We WANT to work with them, because without them, doctors would see a lot less patients everyday. Midlevels are a necessity in this day and age.

 

[loveoforganic2]

2. So let's say we include a stipulation that any consultation of physicians constitutes a "treatment failure" on the part of the DNP instead to get around #1 above

If you included the same stipulation for physicians, that could be a possible method, but it doesn't have great face validity as a quality of care measure (and I doubt there's much research on using it as such). What could be an interesting study would be looking into average complexity of referral between MD and NP.

Also, one barrier to any MD/NP comparison research begun by MD's is that those MD's seem to care about endangering their relationships with their nurse coworkers (concern voiced to me before by a physician anyway)

 

[peteB]

Sorry, that was more of a response to peteB.



I don't care about working WITH PAs or DNPs. If we say we aren't going to work with them, they'll start their own practices. We WANT to work with them, because without them, doctors would see a lot less patients everyday. Midlevels are a necessity in this day and age.

I don't think midlevels are a neccessity. I think we need more doctors. It's not like we can't get more anyway. There are lots of people who are qualified and want to go to medical school, but can't get in because we don't have enough medical school spots. We might have to lower the bar for MD admissions a bit, but these people are still at least the same or better qualified than your average midlevel program applicant. There are lots of FMGs who are MDs and passed their boards but are doing nothing because there aren't enough residency spots. If we wanted to, we could bring in a ton more FMGs too. We could be using these people as actual doctors instead of promoting nurses to work as doctors.

We can create more MDs if we want to, but we're creating DNPs instead. The only benefit is cost savings, and we can make them faster. But if we cut down on the number of years for primary care residency, or reduced the cost of medical school/repaid loans for people going into primary care, both of those benefits would be gained from MDs. Not to mention that DNPs are looking to make MD salary and specialize as much as they can, so where's the benefit?

 

[45408]

In fact, I believe they will probably be doing surgery within a decade or so, just as soon as the AANP can develop residency programs to facilitate this and recruit turn-coat surgeons to "train" the first generation of NPs who will in turn teach the next.

Nope.

Just wait until they start having complications and watch the **** really hit the fan.

I work with a couple of NPs. They do their own thing, only asking me for help if they're not sure about something. I'd say they're at the level of a good 3rd year resident, forever...once things get more complicated or far away from a textbook, they get a little spooked.

I have yet to meet more than one that is anywhere near that level...

They don't complain much, which is more than I can say for the average PGY-3....or hell, even the average intern.

Yeah, they work less and get paid much more. Probably no call, no weekends, no holidays, and twice the salary. Gee, I wonder why...

Oh, and of course they don't complain to the attending, but you're kidding yourself if you don't think they complain amongst each other. Our midlevels bitch and moan pretty hard at times.

You sure about that? I'd bet a lot of money most people will buy the propaganda put out by the AANP.

This will be the line of people waiting for a CABG by a nurse practitioner:

 

[cbrons]

I hope you are right, Prowler.

I think I've been on that ride...

 

[45408]

PAs and CRNAs are doing almost all of the procedures in procedure heavy fields like IR, pain, anesthesia, critical care. Not sure that plan will save you. In fact its probably much easier to do the procedural aspects of rads/ccm than it is to do the thinking parts.

IR? They're letting PAs do CT-guided drain placements, trauma arterial coilings, RFAs, and TACEs?

I don't buy it.

Actually, some hospitalist services utilize NPs to do admissions for them.

I've seen PAs on inpatient surgery services as well.

Yeah, they're quite common, and very supervised.

 

[deleted185747]

I would rather be replaced by a computer that follows an algorithm perfectly each time than a nurse. The algorithm would be created by experts and include the best EBM available.

When the patient feels that the computer program failed then they would come see me and the good old doctor-patient relationship + experience would be what saves the day.

Essentially these lowly trained nurses what think they can be doctors are just following algorithms but not as well as a computer could. Computers would be MUCH cheaper too.

 

[xenotype]

IR? They're letting PAs do CT-guided drain placements, trauma arterial coilings, RFAs, and TACEs?

I don't buy it.

Take it for what it's worth.

Not the more advanced stuff, but certainly bread and butter IR work. I don't know what incentive they have to lie about it in their forum.

PAs are not benign and want everything NPs have.

Evidence

 

[Dash2021]

I'll go ahead and make the same disclaimer Ibn Alnafis did: I'm premed (just got my acceptance from my top pick), so take whatever quantities of NaCl you think my opinion warrants.

I came to this forum to see if there was anything useful I could be doing for the rest of the year to prepare for med school (consensus seems to be: not really), when I stumbled over this post. What problems do you see for Dr.'s in the future was one of the frequent questions during interviews, and my go to response was always primary care physician shortage. My offered solution was to expand role PA's and NP's in primary care. *edit* typed "expand scope of practice". Not what was intended, or said.

My thoughts for this were pretty straight forward, and have mostly been mentioned already: Prohibitively low pay for PCP Dr.'s, low training cost for NP's/PA's, relatively low risk to pt's vs. the cost savings that could be had. I know a lot of people have a hard time swallowing this idea, but it's all but a done deal. The reason we have a primary care shortage in the first place is because almost no doc wants to come out of 12 yrs of school, plus residency, carrying 300+k of debt and have the choice of making next to nothing or working like a dog for the rest of their life.

The issue of pt safety only becomes a problem when NP's are treating outside their knowledge base. As long as NP's are trained to refer early and often when they are out of their depth, there should be little/no issue using them as PCP.

As far as NP/PA's getting expanded acute/surg scope...I don't see it happening on a large scale. Not because I'm fundamentally against it, but because the insurance companies won't let it. Regardless of who's doing acute/surg. the law will require the responsible party to have malpractice insurance. The lack of training will necessarily result in more lawsuits, as it will in primary care. The difference is that the nature of acute/surg. care means the losses will be more severe (i.e. death), ergo they will cost more. When the numbers are crunched the insurance companies will come to the conclusion that NP/PA acute/surg. is more costly, legally speaking, than Dr.'s and charge the NP/PA's a fitting premium. Increase premium means they will need to charge more, charging more removes their competitive adv. against Dr.'s. Sure, there will be outliers who are just surgical savants, but insurance companies make policies on pools of data, and even the savants will have to charge enough to compete with Dr.'s.

TL;DR- No one capapble of being in med school could possibly not know there is a PCP shortage, or that NP's are well placed to relieve that shortage. Outside of PPC, it is unlikely that NP's would be able to compete with Dr.'s, assuming that the difference in care is enough to warrant inflated malpractice cost to NP's (if the diff. in care isn't.......). My 2 cents.

 

[Rendar5]

Prowler just to echo smq, I too have seen the majority of my midlevels have several years experience and operate at the level of at least a solid 2nd year on non-complex cases, at least in EM. I have also seen a couple midlevels operate only at the level of interns but this is pretty rare for me personally. It may simply be a function of the field you're in and the midlevel's experience and the complexity of the cases you're talking about..

 

[cbrons]

. What problems do you see for Dr.'s in the future was one of the frequent questions during interviews, and my go to response was always primary care physician shortage. My offered solution was to expand scope of practice for PA's and NP's.

As far as NP/PA's getting expanded acute/surg scope...I don't see it happening on a large scale. Not because I'm fundamentally against it, but because the insurance companies won't let it. Regardless of who's doing acute/surg. the law will require the responsible party to have malpractice insurance. The lack of training will necessarily result in more lawsuits, as it will in primary care. The difference is that the nature of acute/surg. care means the losses will be more severe (i.e. death), ergo they will cost more.

I'm glad you know a lot about how easily a primary care physician can be replaced by someone with less than half the qualifications. Good for you. And you don't think mistakes in primary care can lead to severe consequences?

 

[Bacchus]

I'm glad you know a lot about how easily a primary care physician can be replaced by someone with less than half the qualifications. Good for you. And you don't think mistakes in primary care can lead to severe consequences?

Indeed, there are multiple ways for your PCP to kill you.

 

[loveoforganic2]

Take it for what it's worth.

Not the more advanced stuff, but certainly bread and butter IR work. I don't know what incentive they have to lie about it in their forum.

PAs are not benign and want everything NPs have.

Evidence

PAs, regardless of their own desires (whatever they may be), fall under the board of medicine

 

[calvnandhobbs68]

I'm glad you know a lot about how easily a primary care physician can be replaced by someone with less than half the qualifications. Good for you. And you don't think mistakes in primary care can lead to severe consequences?

I'd just like to know who exactly was buying this at a medical school interview? Actually, it was probably one of the same doctors training NPs to do surgery...

 

[xenotype]

I'd just like to know who exactly was buying this at a medical school interview? Actually, it was probably one of the same doctors training NPs to do surgery...

It is rather shocking. If I had said the same in my interviews, I likely wouldn't have gotten accepted anywhere. Most academic specialists would not necessarily agree with that view, and a generalist would put a giant red X on your file as soon as you left the room.

 

[Dash2021]

I'm glad you know a lot about how easily a primary care physician can be replaced by someone with less than half the qualifications. Good for you. And you don't think mistakes in primary care can lead to severe consequences?

I'm glad to know you can carry on a civil conversation about a contentious topic. I'm fully aware of the difference in education/training between NP and MD/DO. In fact, I stated that there would be an increase in mistakes if/when NP's take over.

The lack of training will necessarily result in more lawsuits, as it will in primary care.

That isn't the point (sadly). The point is this: will the increase in mistakes cost enough to prohibit NP's doing primary care w/o oversite. Medicine is run as a for-profit business, and if the accountants say the risk < reward, it will be so. Right or wrong. If care quality gets bad enough, the public won't stand for it (again, sadly at the expense of very preventable mistakes). As any other buisness, we are at the mercy of the market, and the market is demanding a cheaper product. This is probably the next experiment we, as a nation, will attempt in order to deliver on that product.

I'm looking toward the future and I am attempting to plan accordingly, as should we all. Do not confuse my stance on the matter as collusion: I certainly wouldn't want to see an unsupported NP as my primary. I certainly wouldn't vote for NP's to be unsupervised if I had any say (statement made during the interview did not suppose NP's being w/o doc supervision, which I think is what most are assuming).

*edit* went back over the post and realized I had said expand scope of practice, though that was not what I meant. It definitely wasn't what I said during the interview. Apologies.

 

[cbrons]

I'm glad to know you can carry on a civil conversation about a contentious topic. I'm fully aware of the difference in education/training between NP and MD/DO. In fact, I stated that there would be an increase in mistakes if/when NP's take over.


That isn't the point (sadly). The point is this: will the increase in mistakes cost enough to prohibit NP's doing primary care w/o oversite. Medicine is run as a for-profit business, and if the accountants say the risk < reward, it will be so. Right or wrong. If care quality gets bad enough, the public won't stand for it (again, sadly at the expense of very preventable mistakes). As any other buisness, we are at the mercy of the market, and the market is demanding a cheaper product. This is probably the next experiment we, as a nation, will attempt in order to deliver on that product.

I'm looking toward the future and I am attempting to plan accordingly, as should we all. Do not confuse my stance on the matter as collusion: I certainly wouldn't want to see an unsupported NP as my primary. I certainly wouldn't vote for NP's to be unsupervised if I had any say (statement made during the interview did not suppose NP's being w/o doc supervision, which I think is what most are assuming).

*edit* went back over the post and realized I had said expand scope of practice, though that was not what I meant. It definitely wasn't what I said during the interview. Apologies.

I would recommend you not make this part of your talking points at your interviews.

 

[Dash2021]

You dont know what youre talking about. "Market" and American medicine shouldnt be in the same sentence. And forgive my "contentiousness." i know as a premed you dont appreciate the difficulty, rigor and depth of med school (yet), so when i hear an ignoramous like you say how replaceable we are, it does ruffle my delicate feathers.

I would recommend you not make this part of your talking points at your interviews.

I wholeheartedly agree with you. Market and medicine shouldn't be in the same sentence. Great we agree. I'm not sure how many times I would need to state that this is my opinion of where medicine is likely to go, not where I would like it to go. But for now I'm done responding to ad hominem attacks.

 

[notbobtrustme]

I would rather be replaced by a computer that follows an algorithm perfectly each time than a nurse. The algorithm would be created by experts and include the best EBM available.

When the patient feels that the computer program failed then they would come see me and the good old doctor-patient relationship + experience would be what saves the day.

Essentially these lowly trained nurses what think they can be doctors are just following algorithms but not as well as a computer could. Computers would be MUCH cheaper too.

Actually, Watson, that IBM supercomputer that can ****ing read anything and understand it is set to enter the medical field in the next 10 years. This is the machine that murdered the 2 dudes who dominated Jeopardy in a game of Jeopardy. It's spending the next 4 years or something like that getting 1 billion charts fed into it's machine along with 150 years of research.

Personally, I'd be more scared of Watson in cognitive heavy fields than DNPs/PAs.

 

[Dash2021]

Actually, Watson, that IBM supercomputer that can ****ing read anything and understand it is set to enter the medical field in the next 10 years. This is the machine that murdered the 2 dudes who dominated Jeopardy in a game of Jeopardy. It's spending the next 4 years or something like that getting 1 billion charts fed into it's machine along with 150 years of research.

Personally, I'd be more scared of Watson in cognitive heavy fields than DNPs/PAs.

+1
All hail our robot overlords.

 

[xenotype]

Actually, Watson, that IBM supercomputer that can ****ing read anything and understand it is set to enter the medical field in the next 10 years. This is the machine that murdered the 2 dudes who dominated Jeopardy in a game of Jeopardy. It's spending the next 4 years or something like that getting 1 billion charts fed into it's machine along with 150 years of research.

Personally, I'd be more scared of Watson in cognitive heavy fields than DNPs/PAs.

I doubt it.

Computer algorithms are only as good as their inputs. If Watson lacks one subtle yet important input, it could end in a completely wrong diagnosis and a huge lawsuit. Just look at Siri, it is worthless.

Planes cannot even fly themselves safely yet from take-off to landing. AI driven cars can barely handle traffic conditions. We are decades if not more than a century away from machines replacing highly skilled labor.

 

[peteB]

If Watson lacks one subtle yet important input, it could end in a completely wrong diagnosis and a huge lawsuit.

I'd bet that the same is probably true for a lot of humans too. You don't think people ever miss subtle points? At least a computer can be programmed to certainly never make a mistake again. You can't do that with people. Newly minted MDs make the same mistakes as their predecessors did at that part of their training. Once a bug is fixed in a piece of software, it's fixed for good. Plus, the computers have the ability to learn from millions of cases worldwide that can be integrated into a single piece of software. Each human provider can only gain the limited experience they get from their own practice environment.

 

[take the shot]

New Acute Care Nurse Practicioner "Fellowship" program --> http://www.mc.vanderbilt.edu/reporter/index.html?ID=12364

Now even fellowship trained CC/Pulm docs can be replaced by NPs.

Unbelievable...so many of the acute care NPs have a chip on their shoulder and want people to think they're just as good as a physician. My husband is a pulm-CC doc who works with NPs on the cardiology and cardiothoracic surgery services. They're all full of themselves and think they can run ventilators and pressors. Then the patient crashes and they are helpless; then they tell the nurses to call my husband. He then has to emergently deal with pt issues that would have been better managed before the patient deteriorated.

Additionally, they want to do all of this is without going through all the blood, sweat, and tears of the rest of us who have suffered through med school. To quote my chief resident from surgery: "if you want to be a captain, go to captain school"

I think their is a place for NP/PAs if they know their limitations. I have worked with several good ones but they tend to be in the primary care arena

 

[Textbookversion]

I doubt it.

Computer algorithms are only as good as their inputs. If Watson lacks one subtle yet important input, it could end in a completely wrong diagnosis and a huge lawsuit. Just look at Siri, it is worthless.

Planes cannot even fly themselves safely yet from take-off to landing. AI driven cars can barely handle traffic conditions. We are decades if not more than a century away from machines replacing highly skilled labor.

Actually just to correct on you on the flying thing, autopilot can do the full journey now. I guess from what my pilot friend said autopilot landing can be a little sketchy but that's about the only problem.

The google cars handle traffic conditions perfectly. They've had the cars driving in traffic for thousands of hours without human intervention. They are legally required to have a driver behind the wheel ready to intervene at any moment, but the laws are changing due to lobbying.

As for IBM's supercomputer putting doctors out of work, not going to happen until we have actual artificial intelligence.

 

[45408]

Take it for what it's worth.

Not the more advanced stuff, but certainly bread and butter IR work. I don't know what incentive they have to lie about it in their forum.

That's completely different from what you just said:

PAs and CRNAs are doing almost all of the procedures in procedure heavy fields like IR, pain, anesthesia, critical care. Not sure that plan will save you. In fact its probably much easier to do the procedural aspects of rads/ccm than it is to do the thinking parts.

I'm still a bit surprised that they're doing quite as much as they claim, but it's certainly a lot less than "nearly everything," because there are much more complex procedures in IR than vascular access, thoracenteses, liver biopsies, etc. I simply don't believe the bullsh-t from the guy who says "an IR PA can do pretty much anything the doc does."

 

[45408]

I'd bet that the same is probably true for a lot of humans too. You don't think people ever miss subtle points? At least a computer can be programmed to certainly never make a mistake again. You can't do that with people. Newly minted MDs make the same mistakes as their predecessors did at that part of their training. Once a bug is fixed in a piece of software, it's fixed for good. Plus, the computers have the ability to learn from millions of cases worldwide that can be integrated into a single piece of software. Each human provider can only gain the limited experience they get from their own practice environment.

The human provider is also drawing on a million years of evolution leading them to perceive non-verbal communication at a level that the computer isn't even close to attaining.


Did you read the link?

"When physicians are asking patient’s their symptoms, we’re analyzing a complex amount of information that is not tangible and cannot be spoken or inputted into an algorithm: Eye contact; Subtle physical movements; How they respond to questions – does their tone change when describing a particular symtom, leading me to believe I’ll uncover more information if I ask more about this; How they smell; How they are sitting; The reaction of family members when the patient responds to a particular question; What they are wearing; Any signs of underlying trauma; and much more."

 

[xenotype]

I'm still a bit surprised that they're doing quite as much as they claim, but it's certainly a lot less than "nearly everything," because there are much more complex procedures in IR than vascular access, thoracenteses, liver biopsies, etc. I simply don't believe the bullsh-t from the guy who says "an IR PA can do pretty much anything the doc does."

As I understand it, the more complex procedures are not done in most private practice IR and are sometimes turfed elsewhere. Most community hospitals need IR for vascular access, thoracenteses, and biopsies. Not TACE. That's a lot of work that can be done by someone working at a much cheaper paycheck.

Watson will absolutely not replace MDs. Computers do not have the AI necessary to make detailed observations, and you cannot possibly put every subtle sign in for every patient. It will do well in non-emergent situations as a decision aid on more complicated cases. It will still take an MD to pick the best decision out of the aid Watson provides, as it gives you several possibilities, and it will still take an MD for Watson to get all of the data necessary to become useful.

 

[peteB]

The human provider is also drawing on a million years of evolution leading them to perceive non-verbal communication at a level that the computer isn't even close to attaining.


Did you read the link?

"When physicians are asking patient's their symptoms, we're analyzing a complex amount of information that is not tangible and cannot be spoken or inputted into an algorithm: Eye contact; Subtle physical movements; How they respond to questions &#8211; does their tone change when describing a particular symtom, leading me to believe I'll uncover more information if I ask more about this; How they smell; How they are sitting; The reaction of family members when the patient responds to a particular question; What they are wearing; Any signs of underlying trauma; and much more."

What you've said is true, but I don't believe that the majority of visits to a physician's office require that level of complex analysis of non-verbal communication. I think in the future the computer will be capable of handling the most common problems without issue. Also, midlevel providers can understand non-verbal cues too. The difference between the midlevel provider and the physician is the depth of knowledge, which can be supplied by the computer.

Correct me if I'm wrong, I'm only a lowly 3rd year, but from what I've seen, most of the time physicians are basically working off of lab values and certain stereotypical findings using algorithms. Computers are extremely good at doing exactly this. I know, because I used to be a programmer. Also, computers have instant access to every algorithm and every piece of information in the medical literature, which no human being can ever have.

I wouldn't doubt computer technology. Just think of how much it has evolved in your lifetime alone.

 

[peteB]

As I understand it, the more complex procedures are not done in most private practice IR and are sometimes turfed elsewhere. Most community hospitals need IR for vascular access, thoracenteses, and biopsies. Not TACE. That's a lot of work that can be done by someone working at a much cheaper paycheck.

Watson will absolutely not replace MDs. Computers do not have the AI necessary to make detailed observations, and you cannot possibly put every subtle sign in for every patient. It will do well in non-emergent situations as a decision aid on more complicated cases. It will still take an MD to pick the best decision out of the aid Watson provides, as it gives you several possibilities, and it will still take an MD for Watson to get all of the data necessary to become useful.

I don't believe Watson is going to replace any provider. But I think that a midlevel provider + Watson is going to be a very good and cost effective option when compared to an MD alone. It will make the gap between the two very small.

 

[xenotype]

What you've said is true, but I don't believe that the majority of visits to a physician's office require that level of complex analysis of non-verbal communication. I think in the future the computer will be capable of handling the most common problems without issue. Also, midlevel providers can understand non-verbal cues too. The difference between the midlevel provider and the physician is the depth of knowledge, which can be supplied by the computer.

Correct me if I'm wrong, I'm only a lowly 3rd year, but from what I've seen, most of the time physicians are basically working off of lab values and certain stereotypical findings using algorithms. Computers are extremely good at doing exactly this. I know, because I used to be a programmer. Also, computers have instant access to every algorithm and every piece of information in the medical literature, which no human being can ever have.

The difference between a physician and a midlevel is a better history/physical. That makes the difference between diagnosing a rare autoimmune disease and simply labeling it depression or fibromyalgia. Watson would never prompt the right questions and a midlevel doesn't think of them either. The whole point of medical school is to go beyond lab values and algorithms. Stereotypical findings won't require a computer to diagnose. They are horses and are just as obvious to a midlevel as they are to a physician. Care will be 15% cheaper and just as good under midlevels for 95% of the population. The <5% with zebras will get hurt badly. Midlevels may pick up on nonverbal cues, but they will not necessarily ask the right questions or spot the small details on the physical that make a difference.

 

[peteB]

The difference between a physician and a midlevel is a better history/physical. That makes the difference between diagnosing a rare autoimmune disease and simply labeling it depression or fibromyalgia. Watson would never prompt the right questions and a midlevel doesn't think of them either.

I don't agree. I think this is very possible with enough development and training of the AI. I know what I'm talking about since I have studied the fundamentals of AI. I don't know how long it will take, but there is nothing theoretically that makes it impossible. The difficulty is in getting quality input data. One way we can solve this is to develop sophisticated sensors to gather more and more data directly from the patient digitally.

Let's also keep in mind that if Watson is deployed as a cloud resource (as is planned), it can train off of millions of patients worldwide on a daily basis. The more data points the AI has to train with the stronger it gets. Eventually it will have seen enough of the rare autoimmune disease to suggest it as a possibility.

A history is just asking questions. With enough training from real physicians, the AI will also know all the questions physicians ask and the exact situations in which they ask them. History-taking is not magic. It is just a very complex set of questions in the format of a large and complicated algorithm, which a computer can be trained to follow.

I have no doubt that this can be done. But again, I do not know how long it will take.

 

[Textbookversion]

It's really pretty dumb to try and replace doctors with something cheaper. Physician salaries are like 8% of healthcare costs, and if we need more doctors all we have to do is open more residencies and the huddled masses of FMG's will come.

Having intelligent, well-educated physicians on the front-lines of healthcare isn't an untenable proposition. DNP is a solution looking for a problem and not finding it.

 

[Amnipotem]

Wouldn't this predominantly affect the practice of family medicine? Most family physicians I've had have been incompetent, and let's be frank, most of the time they just refer people to specialists or prescribe antibiotics. Not to mention that algorithmic software is more accurate at diagnoses. http://www.acphospitalist.org/archives/2011/08/diagnosis.htm

 

[xenotype]

I don't agree. I think this is very possible with enough development and training of the AI. I know what I'm talking about since I have studied the fundamentals of AI. I don't know how long it will take, but there is nothing theoretically that makes it impossible. The difficulty is in getting quality input data. One way we can solve this is to develop sophisticated sensors to gather more and more data directly from the patient digitally.

Let's also keep in mind that if Watson is deployed as a cloud resource (as is planned), it can train off of millions of patients worldwide on a daily basis. The more data points the AI has to train with the stronger it gets. Eventually it will have seen enough of the rare autoimmune disease to suggest it as a possibility.

A history is just asking questions. With enough training from real physicians, the AI will also know all the questions physicians ask and the exact situations in which they ask them. History-taking is not magic. It is just a very complex set of questions in the format of a large and complicated algorithm, which a computer can be trained to follow.

I have no doubt that this can be done. But again, I do not know how long it will take.

An AI that powerful and capable will replace the need for CEOs, financial analysts, generals, and economists. If physician careers are threatened by powerful AI, nearly every career is threatened. Its a stupid argument and not worth worrying about.

 

[peteB]

An AI that powerful and capable will replace the need for CEOs, financial analysts, generals, and economists. If physician careers are threatened by powerful AI, nearly every career is threatened. Its a stupid argument and not worth worrying about.

I'm not worried about it. I was just pointing out that it is possible. You and others said that it wasn't, and it could never happen, and I don't think that's true.

I think there is difference between a physician and many other careers though. Physicians don't need to be creative. They most apply the same standard solutions to the same set of problems over and over again. That's exactly what computers are good at. The standard changes once in a while, but the average physician isn't involved in that. The average physician just takes inputs, finds the best solution using an algorithm they've developed over time, and applies it. This can be done programmatically. Other careers require creative solutions a lot more often.

Eventually, any job that is mostly repetitive application of the same set of solutions to the same set of problems (like a physician's) can be handled by a powerful AI. What an AI will be less likely to handle is jobs that require invention and innovation (or physical labor, obviously). An AI is also going to have a hard time writing a novel, or an article or arguing a debate. An AI wont' be very good at coming up with a great business idea. An AI won't be very good at being a leader or motivating people. An AI also won't be a very good moral judge. Not saying these things are impossible, but a lot farther off than being able to completely mirror the cognitive processing of a physician. Physicians will probably be among the easiest of the professions you mentioned for AIs to impact.

The idea of an Army general or a CEO replaced by a computer is ridiculous to even think about. These are leadership positions that require people to be leaders. However, the idea of getting a diagnosis from a computer is not ridiculous at all, given that most people already are searching the internet and using apps and sites like WebMD to get "diagnosed" online.

 

[cbrons]

I don't agree. I think this is very possible with enough development and training of the AI. I know what I'm talking about since I have studied the fundamentals of AI. I don't know how long it will take, but there is nothing theoretically that makes it impossible. The difficulty is in getting quality input data. One way we can solve this is to develop sophisticated sensors to gather more and more data directly from the patient digitally.

Let's also keep in mind that if Watson is deployed as a cloud resource (as is planned), it can train off of millions of patients worldwide on a daily basis. The more data points the AI has to train with the stronger it gets. Eventually it will have seen enough of the rare autoimmune disease to suggest it as a possibility.

A history is just asking questions. With enough training from real physicians, the AI will also know all the questions physicians ask and the exact situations in which they ask them. History-taking is not magic. It is just a very complex set of questions in the format of a large and complicated algorithm, which a computer can be trained to follow.

I have no doubt that this can be done. But again, I do not know how long it will take.

Please stop trying to derail the thread asimov.

 

[link2swim06]

I'm not worried about it. I was just pointing out that it is possible. You and others said that it wasn't, and it could never happen, and I don't think that's true.

I think there is difference between a physician and many other careers though. Physicians don't need to be creative. They most apply the same standard solutions to the same set of problems over and over again. That's exactly what computers are good at. The standard changes once in a while, but the average physician isn't involved in that. The average physician just takes inputs, finds the best solution using an algorithm they've developed over time, and applies it. This can be done programmatically. Other careers require creative solutions a lot more often.

Eventually, any job that is mostly repetitive application of the same set of solutions to the same set of problems (like a physician's) can be handled by a powerful AI. What an AI will be less likely to handle is jobs that require invention and innovation (or physical labor, obviously). An AI is also going to have a hard time writing a novel, or an article or arguing a debate. An AI wont' be very good at coming up with a great business idea. An AI won't be very good at being a leader or motivating people. An AI also won't be a very good moral judge. Not saying these things are impossible, but a lot farther off than being able to completely mirror the cognitive processing of a physician. Physicians will probably be among the easiest of the professions you mentioned for AIs to impact.

The idea of an Army general or a CEO replaced by a computer is ridiculous to even think about. These are leadership positions that require people to be leaders. However, the idea of getting a diagnosis from a computer is not ridiculous at all, given that most people already are searching the internet and using apps and sites like WebMD to get "diagnosed" online.

That is incorrect. Interpreting objective lab is one thing. This is maybe 5% of physician's job. Yes....AI could replace that task.

However, the AI is going to have to be pretty dam creative for the rest of the job. Explain to me how a "non-creative" AI would deal with these situations that I saw last week on wards:

1) A diabetic lady keeps making unauthorized trips to the vending machine which is making her hyperglycemic. The resources don't exist to have someone watch her room.

2) A guy in for COPD refuses to stop smoking while inpatient and says nicotine patches don't work.

3) A lady left her glasses and hearing aids at home and can't even hear me when I yell as loud as I can.

4) Lady is in status epilepticus with only partial seizures w/ recent hx of encephalopathy. Does the AI put her in phenobarbital coma when everything else doesn't work? Since it is likely she could never be extubated from the coma, does it let the family decide or at least see her first?

5) And how does it interprut psych patients?


Every patient has their own novel problems and presentations. Often they are social problems which need someone to objectively weigh their social situation vs. the best treatment and then come up with a creative solution.

 

[peteB]

That is incorrect. Interpreting objective lab is one thing. This is maybe 5% of physician's job. Yes....AI could replace that task.

However, the AI is going to have to be pretty dam creative for the rest of the job. Explain to me how a "non-creative" AI would deal with these situations that I saw last week on wards:

1) A diabetic lady keeps making unauthorized trips to the vending machine which is making her hyperglycemic. The resources don't exist to have someone watch her room.

2) A guy in for COPD refuses to stop smoking while inpatient and says nicotine patches don't work.

3) A lady left her glasses and hearing aids at home and can't even hear me when I yell as loud as I can.

4) Lady is in status epilepticus with only partial seizures w/ recent hx of encephalopathy. Does the AI put her in phenobarbital coma when everything else doesn't work? Since it is likely she could never be extubated from the coma, does it let the family decide or at least see her first?

5) And how does it interprut psych patients?


Every patient has their own novel problems and presentations. Often they are social problems which need someone to objectively weigh their social situation vs. the best treatment and then come up with a creative solution.

These situations you are describing do not require a physician to handle. These are not issues of medical science, they are "people" issues. Number 4 is an ethical issue. You do not need to go to 4 years of medical school and a residency to deal with stubborn people or lack of resources. A nurse or midlevel could handle these things just as well as any physician, and to be honest I've seen many handle numbers 1 through 3 even better than physicians. Please tell me how an MD education is required to solve any of these issues (except number 5), I would love to hear it.

The AI will supply the portion that makes the MD different from the other providers - the knowledge of medical science. This can close the gap.

 

[VisionaryTics]

These situations you are describing do not require a physician to handle. These are not issues of medical science, they are "people" issues.

Good thing we rarely deal with people.

Let's say all you're saying is true. What can we do about it? Like someone else said, if we have self-improving AIs that can handle the most complex things that humans do (non-linear reasoning), 99% of white collar jobs will be irrelevant.

As much as I love sci-fi, quit derailing the thread with your pointless AI doom-mongering.

 

[cbrons]

As much as I love sci-fi, quit derailing the thread with your pointless AI doom-mongering.

Seriously.

 

[SpecterGT260]

These situations you are describing do not require a physician to handle. These are not issues of medical science, they are "people" issues. Number 4 is an ethical issue. You do not need to go to 4 years of medical school and a residency to deal with stubborn people or lack of resources. A nurse or midlevel could handle these things just as well as any physician, and to be honest I've seen many handle numbers 1 through 3 even better than physicians. Please tell me how an MD education is required to solve any of these issues (except number 5), I would love to hear it.

The AI will supply the portion that makes the MD different from the other providers - the knowledge of medical science. This can close the gap.

MD badges to patients are like garlic soaked crosses to vampires. Obviously a mid level can't effectively block a patient into their room.

 

[peteB]

Please stop trying to derail the thread asimov.

Hey, I'm not the one who brought up the AI in the first place. I was just responding to other people. But I've made my points, so I'm going to end the conversation now.

And since you brought up Asimov, I just want to point out that a lot of science fiction predictions came true. Some of these include: spaceships, the atomic bomb, the submarine, the internet, portable music players, televisions, mobile phones, video chat, GPS, video surveillance, e-books, escalators... I could go on and on.

 

[Dharma]

More Doom-n-Gloom from the mouths of babes (of SDN)...

 

[peteB]

Good thing we rarely deal with people.

Let's say all you're saying is true. What can we do about it? Like someone else said, if we have self-improving AIs that can handle the most complex things that humans do (non-linear reasoning), 99% of white collar jobs will be irrelevant.

As much as I love sci-fi, quit derailing the thread with your pointless AI doom-mongering.

I didn't say physicians don't deal with people. I said that physicians aren't the ONLY ones qualified to deal with people nor are they the best qualified to deal with people... You didn't get the point of my post. The person I responded to said that 95% of a physician's job had to do with handing people issues. I don't think that's true, first of all, and if it is, then 95% of a physician's job can be handled just as well (or better) by someone without an MD who has good people skills. The examples that were pointed out include stopping a diabetic patient from going to a vending machine. You don't need an MD for that.

Anyway, I'm not doom mongering, I was just discussing the possibilities of technology in the future. I already said that I'm not threatened by it or worried about it. I think it's pretty awesome, and I would be happy to see it. I would be happy to see a world where more and more things are handled by machines and humans are free to spend their time and brainpower on even bigger things. How am I doom mongering?

The doom monger is the guy who started this thread in the first place. But I'll bite on your second point. There is no such thing as a general purpose AI. The technology takes a lot of time and effort to develop and has to be custom made for each purpose. It will take a very very long time for AI to impact every job out there. Medicine is arguably the most desirable target right now for this type of technology. There's a huge amount of money in it for the companies involved in developing the technology- more so than any other field they might be interested in targeting.

Again, not doom mongering, just pointing out that there will be awesome technology in the future that will change the way healthcare is delivered. We should be happy that medicine will be among the first of fields to benefit from these advancements.

 

[link2swim06]

These situations you are describing do not require a physician to handle. These are not issues of medical science, they are "people" issues. Number 4 is an ethical issue. You do not need to go to 4 years of medical school and a residency to deal with stubborn people or lack of resources. A nurse or midlevel could handle these things just as well as any physician, and to be honest I've seen many handle numbers 1 through 3 even better than physicians. Please tell me how an MD education is required to solve any of these issues (except number 5), I would love to hear it.

The AI will supply the portion that makes the MD different from the other providers - the knowledge of medical science. This can close the gap.

I hope this doesn't come as an alarm to you, but at least 50% of a physician's job isn't hard/objective diagnosing. Except things like pathology and radiology, a good part of your job is dealing with social issues.

The problem is a social worker doesn't understand the significance of every disease. You never get the patient without their social issues intertwined. You have to make a subjective decision on what would be best for the patient.

Medical school/residency isn't required to accurately diagnosis 90% of walks through the door. You go through all that training so you don't miss the 10% of people with uncommon things/presentations.

Hell...a even a normal nurse could probably diagnosis and treat 70% of the people who walk through the door.


I didn't even ask how an AI would handle all the subjective parts of the patient evaluation. For example, would the AI notice the pain seeker with "terrible leg pain" walking normally past me in the hall to his room? Would the AI understand when patients compare a symptom or make a reference to something in pop-culture? Would the robot be able to deduct what the tone of their voice is implying?

Essentially, yes a robot can replace 100% of what humans do....but you are significantly down playing the subjective aspect to a physician's job.

Are you pre-clinical?

 

[link2swim06]

Oh and one other point.

It seems like if you could build an AI to replace all aspects a physician care for then wouldn't the same level of technology be able to find a cure to every disease known to man and essentially make us immortal? (half-way serious question)...

 

[Lokhtar]

No. Diagnostic AI is very different to the level of control over biology that you're talking about.

Diagnostic AI is a stepped up version of Watson (on Jeopardy) which simply mines data of billions of past diagnosis based on physical and test findings to come up with a probabilistic diagnosis and treatment plan. I think it's very likely within the next decade. Likely it'll be only as a 'consultant' and a 'double checker' at first, and most physicians will be superior. But clearly eventually the computer will be superior, especially for non-standard diseases/presentations (because a computer doesn't forget a nonclassical presentation of a disease that only affects one out of ten million people).

Of course, even then, for the forseeable future, you'll still need very qualified people to take physicals, input findings, think about the social and other issues which may affect disease.