The Medical School Dean Who Says AI in Healthcare Moved Faster Than He Ever Expected

Narrator 0:06

This is Monday Conversation, a PYMNTS Podcast. Karen Webster sits down with the visionaries behind the trends for the stories shaping what's next in payments and commerce. In this episode, PYMNTS CEO Karen Webster sits down with Dr. Marschall Runge, former Dean and CEO of Michigan Medicine, to talk about his optimism about generative AI.

Karen Webster 0:37

Hi, I'm Karen Webster. So here's the question on everyone's mind. Will AI help me live longer? Dr. Marschall Runge wrote a book, The Great Healthcare Disruption, that explores that topic in great detail. Dr. Runge is an expert on the topic as the former dean of the University of Michigan Medical School and the CEO of University of Michigan Medicine. He got his MD from Myamamata Johns Hopkins University. In it, in his book, he suggests that AI is more of an assistant than a replacement, a welcome outcome given the doctor shortages facing patients in this country. In his book, he talks about a healthcare system that's broken in part because it's not accessible to those who may need it the most, and in large part because it focuses on health care and not health. Something he believes that AI can improve dramatically using diagnostic tools that create a patient-specific baseline that can be continuously monitored. He says that making that shift not only is better for the patient, it's better for the doctor and the health ecosystem around all of them, saving time, saving money, and improving outcomes. But here's the question that I wanted to explore with Dr. Runge. Yes, AI can be an assist, but with its great potential to redefine clinical engagement and judgment, can it actually also redefine what it means to be a doctor? Hey, Dr. Runge, thanks so much for joining me today. I'm looking forward to exploring all the great topics in your book, The Great Healthcare Disruption. Fascinating read, and obviously a topic that's on the mind of everyone in my the introduction that uh that I did before talking to you. You know, the question everybody wants to know is, will AI help me live longer? How would you answer that question?

Dr. Marschall Runge 2:18

I would answer that question um in two ways. One is uh I think AI can help us live longer. Uh the second is there's such a huge interest in longevity these days, and who knows what the answers are. Um, but I think AI can help because I think it can help us uh with our health. Uh it can also help with health care, but it can help us with our health.

Karen Webster 2:40

Yeah, that's one of the points you made in the book, which is the the current healthcare system is focused on reactive treating patients when they're not well, with the focus really needing to be more on proactive and health. But that also takes the patient, the consumer, willing to invest in health and having the right tools to do that. I mean, how how do we how do we bridge that gap in your view?

Dr. Marschall Runge 3:05

Well, I think there are several different approaches. Um, one is to make it attractive to patients. How do you make it attractive or people? I mean, we're all people.

Karen Webster 3:13

Yeah, yeah, yeah.

Dr. Marschall Runge 3:14

How do you make it attractive to people? Well, one way is to not make it too complicated, in my opinion. You know, you give you give people this laundry list of things you got to do every day. Uh, most people, myself included, can't do that. Um if you say, here's some principles that I think you can follow that will help you uh maintain health, will assure you that you don't have to go to the hospital when you don't have to go to the hospital. Uh, and if you do have to go to the hospital, you'll be in better shape. And then finally, I think there's a place for financial incentives or disincentives. And uh that's not currently part of how commercial insurance works, um, but it could. And it could be part of some of the CMS programs that are getting rolled out with the emphasis on health. So uh I think there ought to be a financial component to it.

Karen Webster 4:05

Yeah, financial incentives are generally um effective because people want to maximize their uh the amount of money they're getting and minimize the amount of money that they have to pay. So it is a good point, but but there has to be an experiment, right? Because you have to be focused on outcomes. And doesn't that sort of re-architect how we how we today charge people premiums and pay out benefits?

Dr. Marschall Runge 4:30

Yes. Uh and and that is a great point. So um and the question is, well, how quickly can that change? So I I think our payment systems all goofed up. Um our views were developed in the 60s uh as an approach to quantify the time and effort it took for healthcare. Uh since then, I mean, how much has healthcare changed? It's changed tremendously. Yeah, and our views in my mind, and I think in the mind of most people who really think about it, that they're irrelevant. But the trouble is they're easy to work with. Insurance companies can um it's it's something that they can measure productivity of big group practices or academic institutions, they can measure productivity of their physicians by measuring RVUs. And so it's it's gonna be a journey to try to get away from RVUs. But I think if we can do that, and there are health systems in the world, as you know, uh that don't use RVUs. And uh they, for example, I think in the national health system uh in Great Britain, they uh you get paid, and particularly in primary care, you get paid to take care of a certain number of patients. And you get paid the same if the patient's sick all the time or if the patient's well all the time. Uh, and your goal is to help them be more well. Uh, and and at the end of the day, people want to be well. They don't they don't want to be sick all the time. Yeah, right. Right. So, but but I think I think our payment system is a in inhibits that and and we're lagging.

Karen Webster 5:58

I mean, the the the system in the UK, though, I mean, it it does generate a lot of doctor burnout because they have to have they have to take so many cases to make a living. So, I mean, that that's also sort of counterproductive for getting an efficient medical system, healthcare system that does give access to the patients who, in your, in your book, you point out who really may need it the most. They may not have access to the doctors they need. I mean, I know myself, even, you know, it takes a long time to get an appointment with the doctor, even if you have good health insurance.

Dr. Marschall Runge 6:33

Right. Yes, all all good points. And you know, we it's easy to look at both Canada and the UK and say, well, I need to get a knee replacement. And and the answer is come back in nine months schedule.

Karen Webster 6:46

Or or or two years.

Dr. Marschall Runge 6:48

Or two years.

Karen Webster 6:49

And good luck walking in the meantime.

Dr. Marschall Runge 6:52

Right. And even even more uh life-threatening issues like needing a pacemaker, there gets to be a a uh long wait to have a pacemaker. So uh I think it's it's it's gotta be a blend, and and I don't think either of those systems work great. Now there's um uh it goes by variable names, but direct primary care, um, which in one form is concierge care that you have to pay a lot of money for. But Medicare is going to support direct primary care. That means smaller panels, uh, so not as much doctor burnout, uh, and uh uh more emphasis on health. And I I I I believe that if we incorporate AI, so AI can have lots of uses, but uh one of the uses is to around access. So I'll just tell you our journey here at the University of Michigan, but I think it's similar everywhere. Uh we used to have each individual physician's office make make appointments, that was chaos. Uh then we went to a centralized call schedule, uh centralized um uh scheduling center, uh, which we we have RNs who do that, and that's improved it. However, we've done a pilot, and many places have done a pilot where you use AI. And so why would AI be any better? Well, it's better because it can look at every single doctor and it can find out that on Thursdays between 11 and 12, I leave that time open in case somebody needs to come in. But most of the most of the time I don't use that time. And and people have that time scattered through their appointments uh all week long, and so AI, just as it can do everything in an instant, can look and say, Well, Karen, we think you need to see um a doctor in this specialty a minute later, it says, Here are four options for this week. And and those options are they're out there. Uh that would be awesome. That would be awesome. And we we need that because access, I think access is the key, and and it's not just that. I'm sure the same can is true in well, I know the same is true in Boston. Our emergency rooms get totally congested because I'm feeling bad. Uh I can't get a hold of my doctor or doctor's office. I'm thinking, well, am I about to croak? And do I need to go to the that's a medical term? Uh need to go to the emergency room, or is this just like the flu? And we don't have any guidance. And I think there are definitely areas in which 24/7 with Google and Amazon have done well with 24-7, but 24-7 uh responses from AI uh can help direct person. And AI always says when you ask it, it says, We're not a doctor, but it could say, you know, it sounds like your symptoms, Marshall, are such that uh we can arrange an appointment with your doctor tomorrow. I mean, that that is feasible.

Karen Webster 9:46

Well, that's I was I was gonna ask you about that because you talk about big big tech has already shown up, but big AI is showing up too, because people talk to Dr. Dr. GPT and as if they're talking to a doctor. And I mean, I think that's amazing, but it can also be potentially dangerous because it could misdiagnose, right? It could say, don't worry about going to the doctor because you're probably just you had a stressful day at work or whatever, a man that just sort of making a exactly right. But but because doctors don't want to get it wrong, which is part of the reason why there are a lot of tests and a lot of specialists, and let's make sure it's not this, let's rule out all the bad stuff. But how so how do you navigate the I want to be precisely right, which AI can really do, can help a clinician do that without the per the patient taking it on themselves to say, well, I really don't need to go to the doctors because I have Dr. Chet.

Dr. Marschall Runge 10:48

Right. Uh another great question. So here's my view on it. I think that AI uh should not be held to a zero error standard, uh, but it will be. Um first time somebody dies uh based on AI advice, it's gonna be headlines, and there's gonna be a horde of people saying, See, I told you so.

Karen Webster 11:10

Right.

Dr. Marschall Runge 11:11

The example I like using is Waymo. I don't know if you go to San Francisco.

Karen Webster 11:15

Yeah, yeah, yeah.

Dr. Marschall Runge 11:17

Yeah, 100 million miles with no fatal, no fatal accidents. And but you know the first time there's a fatal accident with Waymo, it's gonna be all over the news. Right, right. We can't require something that's just unachievable, right? But I think AI, anything that we're doing medically with AI, uh ought to have to be certified and have guardrails. And at least as we're learning more about it. Now, if if you'd asked me a year ago, do I think AI could function this way, I would have said no way, so far away. But it's moving so quickly.

Karen Webster 11:47

Yeah, it is.

Dr. Marschall Runge 11:48

And um so so I think that uh uh it's possible to to thread that balance as we learn more. Now, I do uh I tell patients and I know myself, I use uh AI uh as my preferred source of medical knowledge. Uh and the reason is that AI thinks broadly. So if you want to go look, uh up to date's a great uh reference, uh so is uh WebMD. But you can't say, hey, I'm a 71, which I am, 71-year-old man who has taken these medications, I have these problems, and now I have this symptom. Um it will give you uh a description. And uh another thing that's kind of interesting about this is more and more physicians are doing this, and uh they are sometimes finding, not often, but sometimes finding AI comes up with a potential diagnosis they hadn't thought about. And so they need to think about it. Uh most of the time it's something they thought about or or maybe doesn't make any sense at all. Um but I think for patients, uh using their favorite um AI source uh is probably a very good thing to do because it's also very well annotated, ask questions, and it's fast. So so I think we're in this middle zone where I'm I'm not the government uh really wants to push AI as an independent provider. Uh I'm not ready for that. And I think maybe I will be ready in a year or two or three, but I I don't think I am now. I know I'm not now, uh, but I'm ready for it to do more.

Karen Webster 13:28

But but if when you think about what it means to be a doctor in the world of AI now, I mean it's you describe it as an assistant, but doesn't it redefine what it means to be a doctor? I mean, think three years, four years, five years into the future with young doctors coming in to practice. You know, I ask my own doctors about this, and they say that they have this debate. The doctors that have all the experience are very valuable because they have all the experience. I've done 2,000 of these surgeries and I can I've seen every possible permutation. AI can say, I can go across 10 million surgeries and I can tell you all the things that might go wrong, and here's how to do for this patient with these things. I mean, where do you see AI being the assistant, but AI completely redefining what doctoring means?

Dr. Marschall Runge 14:23

Um I think uh I think AI can already be a great assistant. And uh it can do lots of things. It can help gather information, put it in a in a very logical format. Uh you you know so much about this, I'm sure you're familiar with ambient AI or the fact that you have AI. And it doesn't just create a transcript for your for your listeners, it it creates a really pretty high grade medical note that you have to look at, make sure it's correct. Um so it can look at problems in that sort of way. And uh so the question is, well, what about somebody like me? So when I trained, you could literally memorize almost everything. Uh that didn't last too long. And a few years later you couldn't, and now you don't have a chance. And so part of what you're alluding to, I think, is how do we train doctors? So and I think doctors have to have a very strong, broad baseline of medical knowledge. They can't know everything, but they need to know enough that when they see things from AI, uh, that they can put it into perspective. So so getting back to uh your your question, uh, I think that uh some years from now, uh we'll be looking at AI in a different way, we'll be thinking about patient encounters in a different way. Uh it's a fact that we have a great doctor shortage, and particularly in primary care. Yeah and we we've done it to ourselves in the United States. Other countries don't have that same problem. So we we um this is a way that we can extend primary care providers. And finally, so so you and I are having a great conversation. I'm enjoying it. Um, I can see your facial expressions, I can understand you know what you're thinking. Uh, if if I were there and you were telling me that you had your right shoulder, you couldn't move it around at all. I could do some uh testing of that. Hey, I can't do any of that. I'm I'm I'm convinced that it's going to really be able to read human emotion. And that that is an important part of your relationship with a physician.

Karen Webster 16:32

Yeah, I now I I I do I do agree with that. Um, but it it is so interesting because it's so powerful and it can it can be more precise because it's I mean, to your point about the access to all the different sources of information, it's vast and it's beyond you know any human's capacity to actually correlate and curate and and actually then produce a produce a produce an outcome or or some sort of recommendation.

Dr. Marschall Runge 17:00

Yeah. Couldn't agree more.

Karen Webster 17:02

Where where do you see the the greatest opportunity for advancement right away using AI? Or where has it already made such such inroads?

Dr. Marschall Runge 17:13

Well, I think it's made great inroads already in uh operational issues. So there are now several examples of hospitals that are AI enabled.

Karen Webster 17:23

Yeah.

Dr. Marschall Runge 17:24

One of my favorite examples has to do with a hospital that's uh in Houston, where uh uh they have had AI, let's say you and I and uh eight other people do surgery in that hospital. It's a small hospital. So it observes us. It's like design thinking. It observes us and it finds that uh when you're done putting your last suture and you're done. It observes that when I'm done putting in my last suture, I'm gonna kind of like worry about it and recheck everything. So when you're done, it can predict that your patient is gonna be in the recovery room in 15 minutes. Uh with me, it's gonna say, hey, it's more like 45 minutes. And why is that important? Well, first of all, it's important for utilization of the operating room. So environmental service can get called, you put in your last stitch, and they're called and say, Hey, you can come on up. Uh the patient's gone and clean it up. And for me, it says, uh, Dr. Slow just finished his case, uh, wait 45 minutes, but come up here. And on the patient side, uh the uh friends or family who are waiting, I don't know if you've had this experience, but you go up to the desk and you say when is my friend's surgery going to be over? And they say, Well, they're they're working on it, it's coming along. And uh AI, based on this on this model, uh AI can tell them uh they'll be in the recovery room in 15 minutes, give them 15 minutes to get settled. You should be able to visit with them in half an hour and you know, go get a cup of coffee, do whatever you want. So this this hospital I'm referring to increased their ability to use operating rooms uh by uh almost 20%. So that's a lot of extra surgeries. So we were talking about access through scheduling appointments, roll that out through all our our health system, uh health systems and think about how much more capacity we can generate uh using AI. So that I think that's a great example. So that's kind of an operational example. Um I think that um one one one uh really interesting approach that some uh medical schools are using is to have medical students or residents there's something called morning report, where you go through and you go around with your attending physician and look at every patient and uh have them go ahead and use AI prior to morning report. So it's it's less of getting uh examined on what you know, but it's more on being able to defend what you learned. And so I don't know if you used AI in this sense. I I actually use it when I'm doing my maintenance of certification. It's it's it's okay to use AI. And it gives excellent uh descriptions for why it would select one treatment over another treatment. And so with a resident or medical student, instead of saying uh, well, I don't know, I was thinking about a new blood pressure pill, uh, and the faculty member saying, Well, why are you thinking about that? to say, well, here's what I think we ought to do, and that's based on what I've done before, which in the old days might be looking up on WebMD or up to date or something, but this is much faster and uh probably more accurate. So it's a new way of learning. We got to figure out how we handle that as well while keeping the humanism in it.

Karen Webster 20:50

Great. It's interesting. Your your operational example, I had surgery a month ago at Mass General, and it was precisely time to the minute, you know, how many minutes I would be in surgery, how many minutes I would be in recovery. And my husband was able to monitor that. So, you know, he knew exactly when to show up and and uh and when to take me, and it was precise. I mean, I just thought that was amazing, you know, X number of minutes and you know, it it it does make you sort of make you know, is this like the Formula One pit stop? I thought it's we'll get to the OR and there's computers and timers and everything is going according to schedule. It's it's really efficient, but it it does show you the extent to which these technologies really do produce better outcomes and a more efficient utilization of the teams who are necessary to be part of that experience.

Dr. Marschall Runge 21:45

I'm impressed with uh what uh MGH is doing. I I was there for uh five years, uh some years ago, and uh it's a great place, but this wouldn't have happened when I was there for your experience at all.

Karen Webster 21:58

Well, I can I can Can be the uh the testimonial that it's actually it's actually working. I mean it's I think it's competitive pressure, but I I do think that it's also you know the expectation of the healthcare systems to improve efficiency, to do outcomes, to raise revenue. Obviously, the more surgeries you can do, the more money you can make.

Dr. Marschall Runge 22:19

Um one other thing I wanted to mention, Karen, is is um I think as we think about adoption of new technology, uh people like yourself who have a broad reach and can think through these things carefully, uh bring great value to the to uh the population. And and the reason I say that is it's uh there's so much on social media, on Reddit is my worst favorite. Um, people just throw their opinions out there. Or influencers who have uh no medical medical knowledge but really have strong opinions and are great speakers. Uh, right. So I think having uh and I'm sure your audience is uh uh views you as a trusted source of knowledge, and I think that having these trusted sources of knowledge can really accelerate how we do and don't use technology.

Karen Webster 23:11

Well, thank you for that. I appreciate that, and I I hope that's true. One final question for you. So you've written the book, you've written other books. What's the next book?

Dr. Marschall Runge 23:21

Oh well, I'm working on two books. Oh, geez. So one is a follow-up, they're uh they're gonna be on a similar topic. I'm I'm further along on one of them, but uh one is gonna be a follow-up of the novel I wrote.

Karen Webster 23:34

Um yeah, that was interesting.

Dr. Marschall Runge 23:36

That was so interesting. Yeah. And uh, but the other is going to be a book that I'm just really getting started on, but is a nonfiction book about um what do we know and what do we not know about this immortality quest. It's so much in the news. Yes. There is there's a lot known, but there's also a lot of just noise out there for influencers and others. And so for me, it's been a great project just reading about all the craziness that's out there in the world.

Karen Webster 24:06

Uh and everyone everyone wants everyone wants to live longer, that's why.

Dr. Marschall Runge 24:11

They want to live longer and they want the silver bullet. They they want the the pain, they don't want to necessarily exercise, watch their sleep, diet, etc. Um, so uh these are early on, uh, but I'm working on both of them. So when I get tired of dealing with facts, I just make up some stuff and put it in the fiction book.

Karen Webster 24:29

I love it. Dr. Runge, what a pleasure. Thanks so much for taking the time to chat. And I look forward to those other books that uh that will come. And hopefully we can stay in touch. Thanks for your time. That would be great. Thank you very much. Thank you. Bye-bye now.

Narrator 24:45

That's it for this episode of the PYMNTS Podcast: The Thinking Behind the Doing. Conversations with the leaders transforming payments, commerce, and the digital economy. Be sure to follow us on Spotify and Apple Podcasts. You can also catch every episode PYMNTS.com/ podcasts. Thanks for listening.