Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode 12: The Future of Medical Education: Guest Brent Stansfield
13 Plays5 years ago
Joe and Rolf talk with Brent Stansfield, who is currently the director of medical education at Wayne State University, about the future of health care and the kinds of value that doctors can provide as artificial intelligence and robotic surgery come of age.
We frame the discussion around the article "Medical Education Must Move From the Information Age to the Age of Artificial Intelligence" by Steven Wartman & Donald Combs.
Special Guest: Brent Stansfield.
Recommended
Transcript
Introduction to Cognation Podcast
00:00:06
Speaker
This is Cognation, the podcast about cognitive psychology, neuroscience, philosophy, technology, the future of the human experience and other stuff we like. It's hosted by me, Rolf Nelson. And me, Joe Hardy. Welcome to the show. Okay. Welcome to the podcast.
Research on Medical Training and AI Integration
00:00:27
Speaker
Uh, today we have joining us, uh, Dr. Brent Stensfield.
00:00:31
Speaker
who is the director of education for graduate medical education at Wayne State University. Prior to that, he was a professor at the University of Michigan Medical School. And some of the research that he's done is on what it is that makes students succeed in medical school and how you measure success and what the best way is to train them. He's done quite a bit on quantitative methods. He's taught quantitative methods courses. And today is gonna talk to us about
00:01:01
Speaker
trends in medical education, where it's going, and especially how medical education and doctors might reconcile themselves with the upcoming age of artificial intelligence, how doctors can work with artificial intelligence to provide the best sort of healthcare. The article that we're basing a little bit of discussion on is from Stephen Wartman and Donald Combs, and it's called, Medical Education Must Move From the Information Age to the Age of Artificial Intelligence.
00:01:31
Speaker
So Brent, welcome to the show and thanks for being here. Yeah, thanks for having me. And I will start you out with a question. So we'll get right into it.
AI's Impact on Healthcare Efficiency
00:01:40
Speaker
So what's the future of medicine and healthcare? And especially, I guess you have to think about what kind of world we should be training our future healthcare providers for. Yeah, that's exactly the problem. So the artificial intelligence is going to come to healthcare almost whether we like it or not.
00:02:01
Speaker
And doctors in general don't really want it to. There's a lot of pressures in medicine to be more efficient and to be faster and to be more accurate and to make fewer errors and to be more cost-effective. And a lot of that comes at sacrificing the time that doctors can spend with patients and the amount of attention that we can give to patients
00:02:29
Speaker
personal concerns and fears and thoughts. And so doctors' visits are getting shorter and shorter, all in the name of efficiency and seeing more patients. And doctors are having to spend more and more time working with medical record systems that have sort of become more and more electronic based over the last decade or so. Even in my career, when I first started, I started at a hospital that was still using paper charts.
00:02:58
Speaker
So you would get ordered tests and do referrals and all this stuff would be faxed from office to office and every patient would have this big sort of binder of paper and that would be your chart and now it's all done on computers which means the doctors have sort of over the last decade or so sort of had to learn how to enter all this stuff into these computer systems which are always changing and are now increasingly becoming smarter and smarter in the sense that they're starting to
00:03:28
Speaker
you know, make decisions, not for the doctor, but the default settings are a sort of powerful driver of changing how healthcare is delivered because, you know, the choice to uncheck the box instead of check a box can have huge economic downstream costs as, you know, as these things scale up, sort of opting in or opting out of various tests changes how often that they're conducted. And so these kinds of changes are happening in healthcare. And in general, doctors find them
00:03:58
Speaker
stressful because it adds another layer of administrative hullabaloo that they have to deal with. We as medical educators don't wrestle with this question enough. From where I sit as a residency program education director, my goal is to make sure that our residents are trained to be good practitioners of medicine.
00:04:28
Speaker
But given the fact that they are going to be practicing medicine for the next 20 or 30 years, it's not quite clear to me what their job is even going to be like with all of these, you know, as as intelligent systems become more and more part of their workflow. And so that's sort of why I bring this question to the table. And I'm really interested in hearing your guy's perspective since you wrestle more regularly with these kinds of issues than we do in medicine and in education.
00:04:58
Speaker
Well, I certainly have noticed that going to the doctor, it does feel like half of the time is spent talking to a doctor while they're punching things into a computer.
Patient-Doctor Dynamics in Digital Age
00:05:08
Speaker
And it also does seem that there's maybe more discussion back and forth between myself and a doctor where they may not know everything there is to know about a certain area of medicine. And they may say, well,
00:05:26
Speaker
you know, I just saw this paper that came out that had something to do with it. Let's take a look at this together. So there does seem to be a shift where the level of expertise invested or the level of expertise that a patient expects from a doctor is more of a partnership than a one way street where the doctor holds all the information. Is that something? There's so much information available online now, right? And then also with the advances in technology,
00:05:56
Speaker
it may be the case that there are things that are coming online that the patient might be available, might have information available even before at the same time that the doctor does. Sure, with WebMD and everything like that, yeah. Yeah, it's becoming a real, listening to the doctor's talk in general, they
00:06:15
Speaker
It is definitely a reality that patients come in having already googled all of their symptoms. And the patient will come in already sort of with a differential diagnosis that's already been whittled away. And oftentimes a patient will have sort of pre-decided what they think they have. And they come to the doctor more just to sort of get a treatment for the thing because the doctor's now- Ask for a certain medicine or- Yeah, exactly.
00:06:45
Speaker
Uh, and you know, we see this stuff, you know, there's advertisements on television. A lot of this stuff, you know, drug companies go straight to the consumer to say, ask your doctor about this particular drug. Uh, these are almost diagnostic commercials. They simply say like, do you feel sad? Do you, are you feeling lonely? Do you not, you know, have trouble talking to people in, uh, in social situations? Well, ask your doctor about, you know, well, butrol or whatever. They're basically telling you, this is how you should interact with the healthcare system.
00:07:15
Speaker
In terms of from the doctor's perspective, it really must affect their priors. When a patient comes in and says, hey, I have this or this. I mean, obviously the doctor's like, wait a second, let's really get into this and figure out what's going on. But at the same time, it must affect how they start to think about it. Yeah, exactly.
Innovations in Medical Training Methods
00:07:35
Speaker
And it's funny too, because we train medical students to do what we call backwards reasoning. So I come into the clinic with back pain,
00:07:44
Speaker
And they will ask, OK, well, when did it start? Were you moving anything heavy? Is it worse when you're lying down? Does it hurt at night or in the day? Where exactly in the back does the pain move around? Is it diffuse or acute? They'll ask all these questions to sort of get a picture of what might be causing it. And then we teach them to sort of make a list. We call differential diagnosis, which is, well, it could be your kidneys, or it could be your muscles, or it could be your bone, or it could be just work stress.
00:08:13
Speaker
It could be psychological. It could be neurological. And then they do the tests and the physical examination to try and rule out as much as they can. And then whatever is left on the table after they've ruled things out, they try, well, maybe let's treat this. Let's treat it with a painkiller. Or maybe we should treat it with a change in diet. Or why don't you try sleeping with a special pillow? Try and start with the low-cost things.
00:08:41
Speaker
and then rule out whatever doesn't work, that's probably not the problem. And that's called backwards reasoning in the sense that you start with a bunch of possible causes and then end up working backwards towards whatever the actual cause is. And one thing that happens in medicine is in all expertise, when people get a lot of experience, they don't need to do this anymore. Like a doctor with many, many years of experience under his belt will walk into a room
00:09:09
Speaker
see a patient and just say, ah, this person is hypertensive. I've seen it so many times. I know how it indicates. I know how it manifests itself. And they end up doing what we call forward reasoning, which is they simply then they have to decide to either prove themselves wrong. But more often than not, they'll simply say, listen to this patient's story, sort of talk about the logistics of a treatment and say, here's what I'm going to do for you. And they'll just kind of go with their gut because their gut is most often correct.
00:09:39
Speaker
But that's the issue here, is that at odds, the experienced clinician is now at odds with these computer-based systems that want them to do the backward reasoning. It's a frustrating experience for an experienced clinician, and then it's also a frustrating experience for the educating, for the trainee, because they will be developing this expertise, but the system somehow won't allow them to use it. So this is interesting too.
00:10:07
Speaker
Your background is in cognitive psychology. So teaching medical students, I mean, it sounds like a lot of this is about figuring out how to teach medical students to reason in a correct way. And I'm sure you must have encountered, or you must have thought about how you can reduce bias in medical students. So all of the kinds of cognitive bias that we see, the Kahneman and Tversky stuff,
00:10:36
Speaker
how you basically how you get medical students and then doctors to think clearly about all of this stuff. And then, of course, presenting this to a patient in the clearest possible way. So this seems like a different sort of skill than doctors had 30 or 40 years ago. Yeah, there's been a lot of attention in the medical field, in medical education specifically, bringing in social scientists. And there's a reason that as a psychology PhD,
00:11:06
Speaker
That was my first job out of graduate school, was working in a medical education program because they were hungry for that kind of perspective. And also, I should say, anthropology and sociology. Medicine is very, very sort of inclusive and wants to hear those kinds of perspectives because they recognize the importance of that aspect. And it's not their expertise, right? So our medical education system really weeds out the humanities
00:11:35
Speaker
And we pick the best physicists and the people who do best in organic chemistry. Those are the kids who end up going to medical school. And then we sort of have to train them on how to communicate with patients and how to be, you know, we have to give them ethics classes. And it's kind of a fascinating exercise. Sometimes I wonder if it would be smarter to do it the other way around, to take the humanists and train them up on the basic sciences.
00:12:04
Speaker
because they may have some of the maybe linguistic skills or humanistic kind of skills. Yes, exactly. But at the same time, you know, communication is a skill like any other. And that's what we work on. We have a simulated patient exercise that we send all our residents through, where we have actors who portray a patient who is just receiving bad news and is very upset or, you know, here's a scenario where
00:12:31
Speaker
the clinic you work at has made an error and this patient was told something that was incorrect and just found out that it wasn't true and is now very angry. And these are communications exercises and we video them and we tell them, you should go watch this video, see how you reacted in this situation, like watch your own body language, watch the way you spoke and watch it with one of the faculty members and get some pointers. It's a skill. It's kind of interesting to see, at first,
00:13:01
Speaker
We do it with our first year residents and our second year residents. And the first year residents generally have a chip on their shoulder and say like, whatever, like I can do this, this is my job, I'm a doctor. And the second years come back and they're like, yeah, this is tough stuff. I see this every day and I've gotten a lot better at it. So how do you teach doctors to become experts in
Technology's Role in Medical Education
00:13:23
Speaker
their field so that they think like an expert? Well, a lot of it's just osmosis. I mean, clearly there's a lot of
00:13:31
Speaker
didactic training. So they, you know, the first, and I should say to, and this is maybe germane to the whole or central to the whole thing, is that it takes a long time to make a doctor, right? So if you're in college and you decide you're going to be a doctor, you need to take a pre-med major full of courses. You need to take your organic chemistry and your cell and molecular biology. And when you get into medical school, it's basically it's graduate school.
00:14:02
Speaker
we in the medical education call those kids undergraduates. They already got their BA or BS and now they're going for their MD and that's a four year program. And so two years of really intensive classroom study and then they have to take a very difficult standardized test called the USMLE step one. And it is the step one because there is a step two and a step three that they'll take later. And that's a lot of
00:14:31
Speaker
basic science. That's when they do that quintessential anatomy course where they have to dissect a body and take huge organ systems courses and foundations of clinical practice. A lot of coursework. In the second two years of medical school, they're going to do what's called clerkships where they basically shadow doctors in clinics and they'll spend a month in internal medicine and then they'll spend a month on obstetrics and gynecology rounds and
00:14:59
Speaker
And that's the second two years. That's when they figure out what they want to specialize in. And you'll see, if you ever go to a teaching hospital, you'll see an unusually young doctor wearing a coat. If the coat only goes to their waist, it's a short coat. They're a medical student. They're a third or fourth year student. And the doctor, your primary doctor will say, is it OK if Dr. Smith watches this? And you have the right to say no, by the way. You always have the right to say no.
00:15:29
Speaker
But if you say yes, then that's a third or fourth year medical student who's just watching and sort of shadowing and learning. And then- It's fine if they're just watching, but I don't wanna be the first patient that that guy or gal cuts open, for example. Someone has to be the first, right? Yeah, someone does have to be the first, exactly. There is a model, I think it's in the paper that I sent out, there's a model in education, in medical education, they say, see one, do one,
00:15:59
Speaker
teach one. And that's literally true. That's how a lot of doctors think if I've seen it done once, then I can do it. And if I've done it once, then I can teach it. And that's how a lot of the skill set moves through the medical education community, which is is a little eye opening when you see it happening. I think also this kind of start looking back to the topic of technology and the adoption of technology. It seems like one of the things that
00:16:29
Speaker
is maybe not emphasized in medical school, but may become more important as time moves on is how to interact with these tools of these machines that doctors are going to be increasingly relying on, whether it be robots or artificial intelligence systems that provide either diagnostics or diagnostic assistance. So actually running those machines starts to become part of the job.
00:17:00
Speaker
Yeah, running them, sort of dealing with them, figuring out how to cooperate with them. And like you said, Rolf, increasingly, and I've noticed this too when I see the doctor, they're wrestling with a computer while they're trying to talk to you, which is itself a distraction. But it's also completely necessary because this is how they document that the visit happened. And when you come in in six months or see another doctor or your problem changes, that doctor is going to look at
00:17:30
Speaker
this medical record, this electronic record to see what happened. As these things become more sophisticated, the wrestling and the distraction I fear might get worse, but I think that the engineers are trying to make it better. And I'm not quite sure how it's going to shake out. Well, I guess if you're spending that much time training a medical student, so however many more than your 10,000 hours to acquire expertise in a
00:17:59
Speaker
in the area of medicine, you really want to make sure that you're utilizing the time of these doctors wisely. So maybe you're in a good position to suggest what the core competency of a doctor is and how they should be spending their time most efficiently. I mean, should you be delegating some of these tasks of documentation? Is it just a waste of doctor's time? They should be spending all of their time making diagnoses or
00:18:28
Speaker
Speaking with patients. What's the what is the best use of a health care provider's time? Yeah, so right now there is a growing need for what we call scribes Which would be simply someone who's there to sort of be the doctor's human computer interface, right? so instead of the doctor typing things in while he or she is talking to you the scribe is doing the computer stuff and It is shown
00:18:56
Speaker
It is pretty demonstrable that having a scribe, paying a person to be a scribe for a doctor allows the doctor to see more patients in a day, which more than pays for the scribe's time. You just have two big problems. One is that you can't really convince the administrators that that's true, that paying more people is per se paying more money. And so the fact that
00:19:26
Speaker
You know, it ends up being a cost saving is sort of hard to convince the administrators. And the second one is worried, like the more people you have in the in the communication chain, the more possibility there is for errors. So right now, the doctor is the one who enters it into the record. It is the doctor's judgment and the liability rests with the doctor who made the decision. Whereas if you do this intermediary system, you know, a transcription error becomes a medical error.
00:19:55
Speaker
and that might be a big problem. And that's where I actually see a lot of this technology going with the voice recognition and sort of intelligent systems that might be able to sort of figure out what a doctor is trying to accomplish and then maybe insert some administrative control over we, the big system thinks that's a mistake, so don't do it. I think that's sort of where
00:20:24
Speaker
a lot of the intelligent systems are gonna end up being, basically trying to be an electronic scribe. Yeah, I mean, there's definitely the component around the electronic medical records and how those are, you know, capped and maintained over time. But there's all, I mean, one of the areas that I've worked in a little bit is computer assisted detection. Yeah. Or computer assisted diagnosis. And one of the things that comes up in that world is the idea of probability.
00:20:53
Speaker
and how much the machine is affecting a doctor's view of the probability of something being either there or not there.
Interpreting Medical Data and Diagnostics
00:21:03
Speaker
And how difficult, what occurs to me in thinking about that is how difficult it is to think about probabilities. Humans don't do well thinking about probabilities. And I wonder how, it seems that maybe doctors also aren't really taught to think about things in terms of it's 80% likely that this person has this problem or that problem more
00:21:23
Speaker
to a point of differential diagnosis, like let's go down the decision tree, the end of the decision tree will be at a definitive diagnosis, we'll know what exactly this patient has. I mean, in the case of artificial intelligence, for example, let's say you had a computer vision system for radiology, and this device detects, say for example, lung cancer. It's a very difficult thing to see on a scan.
00:21:53
Speaker
It's going to give you some indication where tumors may be, but it's ultimately up to the doctor to decide if that is or is not a tumor. And so it's like now you're starting to deal with the probability that this is or isn't a tumor. And I feel like that's a huge challenge to everyone involved. Yeah. Yeah. When I first got into medical education, I remember working with radiologists around exactly this problem.
00:22:20
Speaker
And they are very big proponents of just classic signal detection theory, the stuff we learned in our psychophysics classes. But the big problem is, what's the right answer? Like in a signal detection task, you generally know whether the signal is present or not as the experimenter. And that's how the statistics are built. But with something like, is this a tumor or not, oftentimes you just kind of have to wait
00:22:48
Speaker
until the patient is dead and then you can test it um which is kind of too late which is too late yeah but yeah you feed that system you know feed it back and you can see if you have scans from people you know five or ten years ago and then you know whether that turned into lung cancer or not uh then you can build those predictive systems yeah no for sure it seems as though yeah i guess there were a
00:23:15
Speaker
What's that from a few years ago where recommendations for breast cancer screening were changed because maybe you can just say something about that because I think you'd know more about that than than I would. Well, yeah. So again, just like classic signal detection problems, you you there's two types of errors you can make. Right. You can have the false positive where you say this person has breast cancer and so we are going to treat it when in fact they don't. And you're going to have the false negative where you just sort of miss it.
00:23:44
Speaker
They have cancer, you didn't see it, and it gets worse. And we generally worry about the latter. We're always worried about missing a diagnosis. What a tragedy it is when somebody finds out they have cancer when it's advanced to the point where we can't really do much about it. And so we try to err on the side of detecting things more, which means more tests and more scans.
00:24:10
Speaker
without a consideration of the cost of false positive. Precisely. And those costs aren't just economic. Having a doctor tell you you have cancer is terrible. And you don't really want to put somebody through that experience if they don't have cancer. It also leads away at the credibility of medicine in general, if doctors are handing out false positive diagnoses, if there's a lot of that
00:24:38
Speaker
anecdotally going on, you're going to have fewer people taking doctors seriously. And it just seems impossible to get a patient to to not request a scan when they feel as though there's a possibility that they could learn information that they didn't know from that scan. Exactly. In fact, it's it's almost there's kind of a boutique medicine problem. I haven't heard about it in a while, but a while back, there was a lot of worry because older people of means were just having full body scans.
00:25:07
Speaker
You just go to an MRI, they'll take a complete head-to-toe scan of you and have a radiographer look over and say like, well, this might be a problem, and that might be a problem, and that might be a problem. And you can imagine it's just ripe for false positives. It makes you feel better. Yeah, it's a false positive problem. But I think as the individual, I mean, think about it from the perspective of the system. Sure, there's additional costs that come into play. But if you're the individual, or even if you think about it from the
00:25:36
Speaker
the grandmother test, like what would you, how would you want your grandmother to be treated in a certain situation, right? You know, the, you would want to get the test to see if there was some probability that you may or may not have a certain condition if there was, if you felt there was a real risk there. You know, maybe there's, maybe there's some individual differences on that. I feel like I'm of that mindset too, but you know, people that get
00:26:05
Speaker
say DNA testing, even just the partial DNA testing from 23andMe or something like that. I think there's a range of curiosities about what people want to know about their odds of having a particular disease. And I think a lot of it comes down to, and a challenge really, it gets back to my original point, which is that we have a very difficult time understanding probabilities, right? The idea that, all right, you've got like a 20% greater chance of having, you know, hypertension. Well, that's actually basically nothing.
00:26:36
Speaker
20% higher chances because the base rate is so low. I mean, hypertension is a bad example because the base rate is relatively high, but let's say some really unlikely disease. The lung cancer is a good example, yeah. Exactly. People are very worried about lung cancer and the incidence is not as high as you think and the 20% increase or decrease isn't as big a deal as you think it is. Okay, so just to clarify, so you mean so if it is
00:27:00
Speaker
If you have a 0.1% probability of getting lung cancer, then a 20% increase in that 0.1% means your overall chance is still very, very low. That's right. Exactly. And getting that point across to people is difficult, especially it's like, oh, well, there's a certain percentage chance that this is or is not a tumor. It's something that is not even really expressed, I don't feel like, in the context of medicine.
00:27:29
Speaker
They're just presented as you do or do not have something. That's right. And it's funny that the idea of a diagnostic test being the correct answer is a misunderstanding, I think, that doctors have to kind of fight against. Every test has a sensitivity and a specificity to it. And a hit or a miss isn't the correct answer. You have to sort of weigh that.
00:27:57
Speaker
Right. Absolutely. And, you know, especially when you start to talk about, like, for example, the systems that we were developing, you know, for detecting strokes, you know, would send an alarm or alert depending on if, you know, there was a potential occlusion, large vessel occlusion in the brain detected in the scan from the algorithm. This is like maybe 90 percent sensitive, 90 percent specific.
00:28:25
Speaker
there's still a lot of times it's gonna send you a false negative or send you a false positive or not send you a false negative. But the doctor has to then take into account the fact that this machine is telling them that there's potentially a stroke there, a clot there. And so it must affect their bias and their judgment, but also then of course there's the liability issue that you mentioned before as well. Right, and it gets even more complicated with sort of
00:28:55
Speaker
maybe psychosocial type things like is this person exhibiting signs of depression or am I going to see am I am I seeing signs of somebody who's experiencing domestic abuse right where you see a you see a behavior or a patient says something and then you have to make the decision whether to pull the trigger on this like am I going to set into motion of a bunch of treatment decisions that are really going to have a big impact on this person's life
00:29:23
Speaker
And the diagnostic tests involved are very, they're fuzzy, right? There's a lot of error. So there really are a lot of judgment calls that need to be made. And then on top of that, and to get back to kind of the psychophysics model, doctors have a stat that I'd never seen before in psychology.
Understanding Treatment Effectiveness
00:29:40
Speaker
They call number needed to treat. And number needed to treat is, I can't remember exactly the right ratio, but it's a way of turning a log odds around false positives and false negatives into the number of
00:29:53
Speaker
times you have to say, yes, this person has this thing before you can save one life. Wow, I've never I've never heard of that. That's interesting. Yes. The doctors are actually doing this calculation, not just on a patient to patient level, but they're treating a population. How many times do I have to give this medicine out to people who I suspect have this disorder in order to successfully have one outcome, one positive outcome? So it's a straight out cost benefit analysis. Yeah.
00:30:23
Speaker
saving one life is the, is it saving one life or is it doing good to? It's generally saving one life, but it's, you know, it generalizes to any positive outcome. And number needed to treat is sometimes, you know, 20 or 30 or 40, which means that you're throwing away 29 of your treatments are not saving that person's life. But you just don't know which, you know, which one of the 30 is the one. I think that might be a good place to take a break.
Future of AI and Robotics in Healthcare
00:31:03
Speaker
Okay, so I'm gonna, I just wanna start it out like this. So I noticed on the back cover of my New York Times Magazine today, there's an ad from Mount Sinai Hospital and it says, if you need robotic assisted surgery, consider whom the robot is assisting. That was an interesting and especially relevant thing for this conversation, right? Because the doctor's role here is someone that is there to be trusted and who's kind of the human element to and otherwise,
00:31:33
Speaker
know, electronic and and robotic system. So I wonder what sort of speculations you might have on long term, the long term future of the role of a doctor and of artificial intelligence systems in healthcare. I mean, you know, long, long time scale, ideal world, if everything develops and we get useful kinds of AI, what's the role of the doctor? What kinds of tasks can be
00:32:03
Speaker
algorithmatized and what sorts of what's left for a doctor to do? Yeah, I think in general, like if we go far enough out in the future, you either have the robot apocalypse where health care has just been completely automated and you, the patient, have no say. Algorithms are simply deciding what is best for the population and the society and your health concerns are of no concern.
00:32:33
Speaker
Well, and I think if we're your utility as a person is if we're if we're in an actual hellscape type Robo apocalypse, I think we have there's all kinds of medical torture that these robots can do. Oh, sure. Of course. That's right. It could get even worse than that. Yeah, well, of course. But I mean, you know, like short, I mean, not short, but like let's say intermediate term, you might have a system, for example, where you'd have a machine that integrates like a scanner
00:33:02
Speaker
and a robot. And so like, for example, if you went in for a stroke and the scanner basically identifies that you have a clot in your brain, the robot might be able to put in the catheter, pull out the clot, and you come out of the scanner and your stroke is gone. And maybe there's some human intervention there. Maybe the human is somehow controlling the robot a little bit, or at least telling it to go or not to go, but maybe
00:33:31
Speaker
you know, a human being never actually touches you in any way during that whole process. And I've heard some people say that they feel that's maybe only 10 to 15 years away where something like that could be possible. Yeah, I think surgery is a real growth area for a lot of these intelligent systems. Um, the, the sort of delicate hands, and we already kind of have this sort of these laparoscopic tools where that, you know, you can basically take somebody's appendix out now without cutting them open. You know, you need one,
00:34:02
Speaker
puncture wound to get a camera in there and another puncture wound to get this sort of special multi-fingered tool in there. And that's it. And once you're in with a camera and a tool, it's just a guy playing kind of a video game. This is fascinating instrumentation. And I don't see why that stuff can't be more and more, if not completely automatized, can be more guided by intelligent systems the same way self-driving cars are becoming more
00:34:32
Speaker
sophisticated over time. And the role of the human in that maybe becomes more of a judgment and a decision tool rather than the person who's actually guiding the little movements of the little machine that's in your body. I think there's a lot of growth there. What's interesting to me is that we're fighting against a growing, sorry, we're not really fighting against it. It's coming. And we have to learn how to integrate it.
00:35:01
Speaker
I went to a conference, a medical educators conference, where the plenary speaker was speaking about the real problem of physician depression and physician stress and burnout.
Mental Health and Automation in Medicine
00:35:14
Speaker
We lose about 400 doctors a year to suicide, which is about the size of a large graduating class medical school. Is that worldwide or? That's in America, just in the US. And a lot of that is just the
00:35:30
Speaker
the profession is becoming more and more stressful. And a lot of that is because doctors are feeling that they're losing agency, that they're simply becoming cogs in a big healthcare machine. And a lot of this medical records stuff, electronic medical records stuff we're talking about is part of it, that they are becoming sort of data entry specialists and having less and less control over the kinds of care they can give their patients.
00:35:56
Speaker
And it's not what they signed up for. It's not what they signed up for. And it's also, you know, when you're there, the front lines talking to patients who are upset or scared. Basically, they are the tools of an increasingly sophisticated administrative and computer based scheme to sort of deliver care effectively or inefficiently. And so it's a stressful position to put doctors in.
00:36:25
Speaker
And I worry that on our road to the robot apocalypse, we marginalized doctors more and more to the point where we sort of break the system down. And at this particular conference, so that was the plenary speaker, later in the conference was the team, the IBM team, who were there with the medical version of Watson, and they really wanted to sort of push this. They had a different view of things, I'm guessing. Well, yeah, it was interesting because they were selling
00:36:55
Speaker
Watson with a name and referring to Watson as in sort of anthropomorphic terms and saying Watson can help you. Watson is an assistant. Watson is a Watson is not here to hurt you. There's nothing to be afraid of. But of course, Watson is not a person. Watson is just a large algorithm and I don't know why IBM is making the choice instead of all they have to do is is treat this algorithm as a decision making tool.
00:37:24
Speaker
that a physician can use to make better decisions, but instead they're sort of trying to sell it as a sort of a fake person, a fake doctor. So would the view of IBM do you think be that tools like Watson can alleviate some of the, all of these burdens that healthcare providers have to go through, all of the stuff that doctors have to do? Watson can kind of lighten the load. I think that's the basic
00:37:53
Speaker
the basic logic behind it. But I worry, and again, and I don't mean to dog on IBM, they're doing great work. And it's amazing what these decision tools can do. It's I just worry that the salesmanship around it and treating these and even just calling them intelligent systems or artificial intelligence, I think does a disservice to actual human intelligence in that, you know, these are tools and we are building them and they can be useful or they might not be useful. But instead,
00:38:23
Speaker
What what is kind of happening is that we're assuming that they will be useful and we're going to sort of edge out the humans who've been doing this work in order to make more room for these systems that we assume will take over. And I'm not sure if that's the best thing for the, you know, for professions such as health care, at least in the near term. Yeah, I mean, I think for Watson in particular, you know, they're selling it.
AI Startups vs. Giants in Healthcare
00:38:51
Speaker
as a thing, but it's not a thing. It's not even an algorithm. Watson is a program of, uh, it's a commercial program and it's many, many algorithms. It's many, many programs. I also feel like this from, from like a commercial standpoint, from like a technical standpoint, IBM is trying to boil the ocean rather than tackling a specific problem and solving it really, really well and providing a tool that works great for one situation. They see that that,
00:39:20
Speaker
approach is not really big enough in some way for them to move the needle. So they need to basically solve all of healthcare. And they're trying to do that with their Watson brand, which is really what it is. It's just a brand. And rather than being really good at solving a particular problem, which is why so many startups are having success solving one or more smaller problem, whether it be in medical records or, you know, a transcription or,
00:39:48
Speaker
computer assisted diagnosis or computer assisted detection, all these different areas where really you see the best tools and technologies, a lot of them are being built by smaller startups because they're tackling one problem at a time rather than trying to solve all of healthcare. And to your point, I mean, they're thinking about them more as specific tools that a person can use to help them in their job versus trying to basically take over the job of the doctor, which is in some ways what I feel that they're trying to do.
00:40:18
Speaker
It relates to the idea of how much we are willing to trust these machines and at what stage. I feel like this gets back to the idea of probability. We're not thinking in terms of probability. How likely is that this machine is correct and using it as a tool that we can use to help and aid our judgment, which is where it should be because we're assuming that the machine is right and we're not exposing this probabilistic sense enough, I feel, so that we're not
00:40:46
Speaker
distrusting the machine enough. Yeah, I agree. And it's maybe even a larger point of we don't distrust a lot of these tests enough, like to get back to the question of, you know, should we change our breast cancer screening program to produce fewer false positives? And then the great social backlash against that, because by having fewer false positives, we have more false negatives. But what is the state of the art right now in
00:41:14
Speaker
what medical students are exposed to in terms of artificial intelligence? Very, very little, as far as I can tell, because it doesn't really exist yet, at least not in a way that we can say this is what we're going to be doing with our artificial intelligence tools. Electronic medical health records are here. They've been implemented. They're pretty much everywhere. I'm not aware of any place that still uses paper. In fact, I believe the Affordable Care Act
00:41:44
Speaker
sort of mandated a move away from paper. We've done this to ourselves in a lot of ways. But yeah, in terms of education, it's not clear what we should be teaching our students because we don't know, especially medical students, because they still got three or four years of residency ahead of them. I don't know. I don't know what we're going to be using these tools for in five years. I found so the paper that we use to start discussion with medical education must move from the information age to the age of artificial intelligence.
00:42:15
Speaker
I wasn't sure what to make of the claim that they started the paper with saying that the information age was something that ran roughly from the 1970s to the 2010s when machine learning came around. And now we're in the age of artificial intelligence. And I thought, is this a generally accepted idea that we're outside of the information age and something has surpassed it and artificial intelligence is
00:42:43
Speaker
is genuinely taking over and I thought that seemed a little premature. Yeah, I'm with you exactly and that's kind of where I sit in this decision as well. I worry that the inevitability of this stuff is assumed and that the format of it, again just sort of using the word intelligence rather than, I don't know what, maybe there's a better term you guys can help
00:43:09
Speaker
Well, I don't know what intelligence is. Yeah, we don't know what intelligence is. Yeah, exactly. Artificial or otherwise. Right, exactly. But yeah, it's going to take a toll on the industry as we just kind of assume that it's going to take over. Well, I think this question of agency is interesting that you bring up in the health and well-being of the doctor because it's going to affect, it is affecting everyone, right? I mean, it's partly the machines, but it's partly just
00:43:39
Speaker
the societal machine of the way that the healthcare system is set up, right? Where doctors are becoming cogs in this factory, right? They become like assembly line workers in a factory. And that's in some way, the design of the healthcare system is incentivized to move in that direction. And it's dispiriting to doctors. And I wonder- So maybe there's, oh, I'm sorry, go ahead, Jeff.
00:44:09
Speaker
No, no, I was just gonna say, I wonder, is there a way to improve upon that where we can take the human beings involved in the system as an important component? Yeah, and maybe even there's a way to use these intelligent systems to promote physician well-being. I wonder if there's a use of them, turn it on its head instead of like turning doctors into cogs in a machine, elevate doctors.
00:44:39
Speaker
What is it about doctors that we can leverage so that they're spending less of their time being data entry technicians and more of their time being the humanists that they want to be, that they trained to be? You see an awful lot of students going through your program and they probably have a conception of what it is that they're getting into when they start out at medical school. So what are the reasons that
00:45:06
Speaker
people are getting into medicine and what sort of misapprehensions might they have? Well, um, it's, it's a real, it's a real pro social thing to do with your life. It's an amazing, the impact you have on people's lives as a doctor is, is all for the good. And, um, my hat's off to anybody who, who goes into medicine as a, as a profession, because it's really, it takes a lot of personal sacrifice.
00:45:33
Speaker
It takes a lot of hard work and a lot of dedication, and you can do a lot of good. And the students that I see, both the medical students who come into medical school and then the residents when they get out, and you're sort of a doctor with training wheels for three years, that's the residency system, they're dedicated, they're smart, they're fun to work with, and I love them all to pieces. I think, however,
00:46:01
Speaker
that we still have a societal myth that being a doctor makes you wealthy and a lot of these suits. Is that not true? Well, it can be true and it can also not be true. A lot of these, a lot of the residents that I see are several hundred thousands of dollars or even million of dollars worth in debt from having to come to medical school in their residency program. They're going to be making maybe 60, $60,000 a year on average.
00:46:29
Speaker
for three years, working 80-hour weeks. A lot of them are married and have kids, and this is their career. There's really no upward mobility except to finish your residency. And then after that, some people can parlay that into a very lucrative career. But the primary care, the hours don't get shorter, and the compensation doesn't get a lot better. It can. It depends on who you work for. But if you want to start
00:46:59
Speaker
your own clinic. If you want to be the frontline person who is doing good in the world, you start a clinic and you have patients that you can spend time with, you're not going to be able to make huge amounts of money. So if there's a huge amount of delay of gratification that has to go on in a medical student, someone who's pursuing this for the amount of time that it takes to develop a doctor, what are the odds that
00:47:28
Speaker
You know, they end up in some place where they feel gratified by all of that, and they're in a position where they can leverage everything that they've actually, everything that they've been taught and everything that they've come to expect. What are the odds that that happens? I suppose that's where you see some of that suicide rate. That's right. Yep, and the burnout. And that's why the odds are too low. It should be higher. And I feel like in the medical education system, and I think there's a lot of movement toward this,
00:47:58
Speaker
and you see it in that paper that we're using as our jumping off point, that there's a real desire to push leadership training and team building and communication skills to sort of empower our trainees who will be future doctors so that they're in a better position to advocate for themselves and build healthcare systems that are more effective for patients and be more central in the process since it seems like the march of machines is going to be taking over
00:48:28
Speaker
a lot of the work that doctors have been doing for the last 100 years. Well, if we've got a robo-pocalypse, what is the robo-utopia that might exist in 20, 30, 50, 100 years if things could be improved for the welfare of doctors and for patients with the assistance of artificial intelligence? Well, I'll give you an example. So I heard a community practitioner out here in Michigan on the western edge in the lakes area
00:48:59
Speaker
has a clinic and what he does when he sees a patient, especially an older patient who may or may not have a family member with him or her at the time, he gives the patient an iPad, he himself has an iPad, and then during the clinic visit at any point that something interesting is said or a question is asked or important information is being either being talked about
00:49:27
Speaker
Either of them can push a button on the iPad. And what that does is it puts a little timestamp. At the end of the clinic visit, the patient has, and the doctor both, have access to a video of the interaction with all these little timestamps. Who pushed the button? The doctor flagged this as important. The patient flagged this as important. And that can be shared with family members. And it can be, it's part of the medical record. The doctor can go back and review. And that way, this is a kind of tool where
00:49:57
Speaker
It's patient centered in the sense that we're focusing on the conversation that you're having with your provider and the documentation is direct. Did we talk about this? What exactly did the doctor tell me to do when I came home? Which is a big issue. There was a big study at the University of Michigan a few years back where they just interviewed patients leaving the doctor's office and they simply asked, what did the doctor tell you to do?
00:50:23
Speaker
And it was, I can't remember what the exact percentage was, but it was not a hundred percent. I love the, I love the idea of documenting it like that and being able to timestamp. And I, and I think that as systems get more intelligent, you can really leverage that kind of system so that, you know, medicines are dosed correctly, that you could connect your pharmacist to that kind of thing. So that if your doctor says you need to get this prescription,
00:50:52
Speaker
it gets packaged and sent to your house so you don't have to go to the pharmacist. Stuff like that, and I think a lot of things could move in that direction, sort of facilitating the patient-doctor communication, facilitating the documentation of what was talked about in the next steps, and facilitating the patients, empowering patients to follow their treatment decisions. I like that. I mean, it also kind of speaks to the issue of what we were talking about before around
00:51:22
Speaker
accessibility and making it more equal in terms of how people can access healthcare and doctors can use the latest and greatest technologies. If these types of technologies are available everywhere and doctors and patients can communicate more effectively both together and then remotely, it feels like there's an opportunity there to start pushing out the highest quality of healthcare to the
00:51:50
Speaker
farthest away and most distant clinics from the center of, you know, you can get the best healthcare, not just in the top medical schools, but also like, you know, in the rural clinics and other places where you might not have access to that today, but maybe in the future we can use technology to actually reduce inequality in the delivery of care. Yeah, I agree. I mean, I think all physicians want to do the best for their patients.
00:52:19
Speaker
goes without saying and without question. But the dissemination of information, we're finding that this works better, or this is the best dosage of this drug, or these are the treatment conditions that need to be met for this disorder. Disseminating that out from the big research hubs into all the various world and far-flung clinics is an undertaking. There is a big continuing medical education system that
00:52:49
Speaker
attempts to do that, but it is slow going. And yeah, I see there is a big role for using technology to make those innovations go farther and faster. What kind of advice do you think you'd have to a medical student, maybe someone who's in college thinking about medical school, about what kind of future they might be looking at? Oh, I would definitely go. Go talk to doctors. Have to go and see it. See it for yourself.
00:53:20
Speaker
It's really kind of amazing when you train someone to become a musician, you give them an instrument from day one and they are playing. That's what the lessons are. They are, here's your instrument. And maybe when you go to the conservatory to become a professional, that's when you take your theory classes. And that's when you take your music history classes. Doctrine is the other way around. We do all of the sort of knowledge work up front. And then, and only then, are you allowed to actually go into a clinic and wear a white coat and see what's going on.
00:53:50
Speaker
So it's a very different feeling being in a clinical setting, not as a patient. So if you're interested in going into healthcare, I would highly recommend go see if you maybe get a job at a clinic as a receptionist and sort of see the way it works from the inside. Do you have any programs at Wayne State that help facilitate younger people or early
00:54:19
Speaker
early medical career people for something like this? One of my favorite things that happens at the Wayne State School of Medicine. So Wayne State's in Detroit and Detroit has a lot of healthcare disparities issues because there's a lot of poverty and it's an interesting urban environment. It's a challenging urban environment because there's a lot of people who have large healthcare problems and not a lot of money. And we have a program.
00:54:46
Speaker
that we call the Urban Medicine Program, where first and second year medical students are given a backpack full of just sort of basic first aid and basic medical care equipment. And they go out and talk to homeless people and say, ask them questions. How are you? How are you feeling? Do you have any health care concerns? Are there any anything you want me to look at? Do you have any wounds that need treatment? Have you been diagnosed with diabetes? You want me to check your feet? And I think that that
00:55:17
Speaker
is the kind of experience that a lot of younger medical students and even a lot of people who might be interested in going into medicine should sort of get in the habit of learning how to talk to people, learning how to talk about healthcare issues with people, which is a little daunting at first, and to sort of become a hero, getting used to becoming kind of that somebody's health hero and becoming comfortable in that role, because that's the tricky part.
00:55:47
Speaker
on and we don't teach it early enough. Well, Brent, I think that's really maybe a good place to wrap up the conversation here. I think that the work that you're doing and improving the education of doctors is super important and really fascinating stuff. So I appreciate your coming on the show and talking with us today. No, thank you so much for having me. This has been really fun.
00:56:11
Speaker
Yeah, and I think we should talk every five years for the next hundred years and then we can see what the progress of artificial intelligence looks like.