Episode 757: Artificial Intelligence Impact on Health Care

Published Sep 29, 2024, 9:00 AM

Newt discusses the transformative impact of artificial intelligence (AI) on healthcare with Dr. Marschall S. Runge, M.D., Executive Vice President for Medical Affairs and Dean of the Medical School at the University of Michigan. Dr. Runge shares insights on how AI is revolutionizing drug discovery, medical diagnostics, and healthcare management. AI's ability to analyze vast datasets is accelerating the development of new treatments and improving early diagnosis of diseases like Alzheimer's and diabetes. Dr. Runge highlights the potential for AI to reduce healthcare costs by minimizing unnecessary testing and enhancing the efficiency of medical records management. The conversation also touches on the challenges of integrating AI into healthcare systems, including regulatory hurdles and the need for collaboration with insurance companies. Additionally, Dr. Runge discusses his novel "Coded to Kill," which explores the vulnerabilities of electronic health records to hacking.

On this episode of News World. How is artificial intelligence going to impact healthcare? Many experts have indicated AI may change the way we diagnose disease, manage medical records, and review medical data. High powered artificial intelligence and machine learning are enabling scientists and doctors to create and analyze vast amounts of data and to develop new molecularities to disrupt targeted diseases. My guest today believes we are in the cost of really significant medical breakthroughs. I'm really pleased to welcome my guest, doctor Marshall Rangey. He is the former executive Dean of the University of North Carolina School of Medicine, is currently Executive vice president for Medical Affairs and Dean of the Medical School for the University of Michigan, and his most recent book is a techno medical thriller entitled Coded to Kill. Marshall, thank you for joining me. I think this is very cool.

It's a privilege to be on your show. Thank you very much.

Was it a big shock moving from UNC to Michigan.

Well, I had to change sports. You'd see all basketball, no football. Here is Maine, all football, no basketball, although there's been some good teams.

Although Michigan at times has had very good men's basketball teams. I was just curious. I thought that'd be almost a culture shock. Now, in a piece you published in Real Clear Health entitled We're on the Cusp of a Historic Epoch of Discovery, you describe the medical breakthroughs that we're experiencing right now. Will you discuss some of them?

Sure?

The AI movement's been termed a Promethean moment, meaning you know everything's going to change about everything. I tend to be more or less a believer of that. I'll just give you a couple of examples. One has to do you mentioned with drug discovery, and in the past, if you had a great target that looked like it was going to cure or significantly positively impacted the disease, you'd get one hundred medicinal chemists working on creating a molecule, and that process itself much less testing would take years. Now, and I'm familiar with a drug discovery project which was done using AI, taking a big database of drugs, loading training the AI algorithm, and over about three days it identified about five molecules to treat a disease that was virtually untargetable today, and all five of these turned out to be have potency and a couple of more compounds unknown compounds. So it's almost like thinking back to natural products, where you dig something up out of the soil, you find a new antibiotic. So it finds things that I find incomprehensible how it brings all that data together, but it really does. And so if you think drug discovery, I think that's awfully important because when we think about healthcare, we have so many diseases that are difficult to treat. Some are rare diseases, rare genetic diseases. I mean, one of the biggest problems is type two diabetes, are Alzheimer's and these are very difficult to treat diseases that I think diabetes has gotten better, but I think we'll have solutions for in the coming few years.

When you say a few years, are we talking a decade, two decades and what are you thinking?

I'm thinking less than that, less than five years, maybe as soon as two or three years. Because another approach to drug discovery, which can be aided tremendously by AI is using RNA therapies, inhibitory RNA therapies, and they also can be developed very quickly in a very targeted way.

It's just kind of stress our regulatory system.

For sure, in a lot of different ways. So as you, I'm sure know much more than I do about it. Some of these new therapies are enormously expensive, so how can we afford them? And that's where I'd like to flip over to thinking about AI and healthcare. We have so far to go in terms of reducing the cost of healthcare, and we desperately need to do that. So there are currently some small companies out there that enable the use of an AI bought a one more place a doctor or a nurse, but is available twenty four to seven that can get far along in diagnosis. And then if you take that along with some new technologies. There was a I thought a fascinating paper in the Washington Post about Alzheimer's disease where five different groups describe five different ways of early diagnosis of Alzheimer's, recognizing patterns that as people. As humans, the most brilliant humans can't recognize these patterns because they bring to other so much data. So I think if you combine some of that, which leads to earlier treatment for chronic diseases. The payback won't be overnight, but the payback for the healthcare system of the United States could be over a decade. We could really see a dramatic reduction in what I'll call unnecessary testing. Maybe it's necessary today, but it won't be necessary in a couple of years.

Yeah. I had the experience. I went in to you have my teeth cleaned last week, and the dentist now has an artificial intelligence component. When they look at the X rays, the artificial intelligence component scans the same data and can identify at about ten times the accuracy what the dentists can identify. It's a local dentist, not some of the advanced research center. They said, it's just a technology that's come online, and that it dramatically enhances his ability to care, for example, very very early if you have a cavity. You've got a couple of things you mentioned, and one of them is this whole question of really advanced imaging techniques that I think are going to revolutionize our ability to have early intervention, which sometimes increases cost, sometimes decreases, but it almost always extends.

Life, right. Absolutely.

Yeah, if you think of any chronic disease, the earlier you're inter being, the better the outcome, So I couldn't agree more. I'll tell you just this sort of a weird research study I read some time ago recently. Really, I had to do with looking at the concept that the color of your tongue can predict disease, and this is like ancient Chinese medicine. But the group that was investigating it looked at using AI in that circumstance, and AI was fantastic, and there was a small group. You know, we're not ready to get AI o our tongues right now, but it proved the point that AI can pick up things we just don't see. But also we don't always know what AI is looking at. So maybe it's not looking at the color at all. Maybe it's looking at the papules and the distribution of the papules of the time. I don't really know, And that's one of the great mysteries of AI. You don't know what all data brought in, but it sure does turn out to be provocatively accurate.

How much research do you see long on both Uncn and Michigan in the application of AI.

I don't follow things that you and see quite as closely as I was, but I can tell you we have an explosion of it. So we're looking at applications of AI direct applications in healthcare. We're using AI to be able to organize and scan and make much quicker decisions about electronic medical records. I don't know if you've ever requested your own electronic medical record, but you know it's like page after page after page after page. And AI does a fantastic job of just summarizing that and summarizing maybe three hundred pages down to two or three pages. And for your care provider, then the main thing I think that helps is they don't miss stuff. I know when I start looking through a medical record, after a while, I'm just not en off. So it makes it more accurate and much easier for the providers. The same thing in terms of thinking about how we interact with what we call the inbox, an epic where people ask a lot of questions. We can't be sure that it's going to answer every one of those, but if you get like an excellent first draft and so it makes it really speeds things up and getting back to patients in real time.

So I think those are great.

But more importantly we're looking at applications of AI in areas that do involve imaging, so like I exams ophthalmology in interpreting mammograms and interpreting MRIs and cts. When a radiologist looks at an MRI, you may have seen them doing this in real time, but they're looking at thirty or forty images. And one thing AI can certainly do is say images one through thirty two, don't worry about their normal image, thirty two, three, and thirty four you better take a quick look at and then the radiologist has much more time to ponder it and try to put that together with the clinical scenario. But I think the next step is going to be AI looking at the medical record and AI looking into the images and pulling it all together.

And I think that's just around the corner.

You know. One of the problems is political. We were looking at a little company that had developed a computerized eye exam. The normal model is you go see your optometrist or ophthalmologists every two years in order to have an exam so they can reorder whatever you might need, and then you go back to the alternating year just to have a quick checkup. Well, they have developed a computerized scam that would perform the second year, and in state after state, the ophthalmology and optometry lobbyists would get it out lawed. I can't believe it, Yeah, because they saw it as a direct threat to their income. One of the great lessons for me when I stepped down speaker, I took two areas, national security and healthcare, which were both very complicated. Healthcare, I think is about ten times more complicated than national security. It's unbelievably dense. But one of the things I learned that was I think a real surprise to me was the degree to which the systems operate in an entirely human manner. That is, if you're a hospital, your interest is optimizing the hospital, and if you're the local pharmacy, your interest is optimizing the pharmacy. And in the process, if that means that you're suboptimizing the country, that's okay for you because you know your pharmacy is doing well. So it made me realize how you both had to figure out an improvement, but then you had to figure out how to get through all these mine fields to get the improvement both culturally adopted. You know, would professionals be wing to do it and to get it adopted by people who might feel that their income, or their prestige or their status was threatened. I don't know how much of that you experience in a medical school, but it's been a fascinating challenge to me.

That's so interesting because it's absolutely true. And I'm currently working on a book with Forbes. Forbes will ask CEOs and different areas, right books and so ours is about changes in healthcare, really disruptors in healthcare. And so all these disruptors are companies like what you talked about, that are coming in and saying, hey, there's a problem, there's a pain point in medicine. We can fix it, and they do. But then everyone's competing with everybody for that same dollar.

And it's true.

You think about hospitals, to think about insurance companies, you think about pharmacy benefit managers. They're all after that same dollar and they're pretty successful, which I think does lead to an awful lot of cost in healthcare. So I can't tell you how happy I meant to hear that somebody who has been is integally involved in decision making at the national level understands this. It's a pleasant surprise to me to.

Go back to artificial intelligence, to what extent is the University of Michigan Hospital using this and trying to integrate it into either accelerating speed or maximizing the ability to deal with complexity.

We're currently I would say a year ago we were using no AI. I'd say we're using AI and ten percent of what we do now. I believe that in the next year or two we'll be using it in half of what we do. Because people are a little anxious about to start with, but I think it's proven that it has real value and is not just a spectacle.

It offers real value.

Is the early diagnostics or the ability to radically enhance research or to create connectivity for the patients. I've always thought, if you think about the number of times in your life you'll go into a doctor or a dentist or something and sit down and fill out paper. And I'm now old enough when they say, you know, list all of your drugs, Well, I carry all my drugs in my phone, so I can pull up my phone and do it. But I think back to my mother, who did not have a phone at that time that was smart. She had a ruggar phone and she's taking and out of drugs. I am certain there are occasions where she went to see a doctor or a hospital and only gave them half the drugs she was using because she just forgot. Shouldn't there be some way to almost automate that whole process.

I think there is, and I think, particularly now that we have electronic medical records that can connect to pharmacy records, we do already look at pharmacy records when somebody says, yes, I haven't missed by dose of lipator in three years, you see they also haven't refilled their prescription in the last two years. So something doesn't quite connect there. I think as we get more and more connected now, I do think that there's a danger of getting connections. I'll come back to that in just a minute. But it's long been a dogma that, wow, if we could just pick medications or surgical procedures or whatever therapies that were targeted toward the individual, rather than saying, well, we know this approach works in sixty percent of people, and many people take high pretension as an easy example. They may try two or three or four drugs before they find one that works. I think by combining when it's going to be possible to combine genotype and other O mix so to speak, the proteomics and others along with medical history to be much more accurate in predicting the kind of therapies that will.

Be really useful.

So that's where I see us going, and that saves time, it's more effective, and it saves money in the long run. When I said I think there's a worry about all this connectivity, that's actually one of the streams of my novel, because once you're totally connected, you're totally connected. And all it takes is one kind of rogue twenty year old at Amazon to know that. G If I put together Marshall's purchasing his medications, his EHR, what can I do with that? And in my novel, it's what evil can I do with it?

Given your prestige and your background and in health, what led you to write a novel?

Well, a couple of things. So I started because I had a patient. Mccardial just had a patient that came in and he could have been a perfect character in a Grisham novel. A attorney worked for the FBI, got on drugs, lost his family and was kind of working his way back.

And he had heart problems.

So one day he said, hey, you know John Grisham, and I'm like, no, I don't think I do. And he said he's an author, and I said, oh. He said, I'll bring you a book. So I read a Grisham book and I was forever hooked on Grisham.

I love the guy.

I loved these thrillers, and I thought, well, it'd be fun to try to write one. I bet I can do that. Of course, that was like fifteen years before I finished my novel. But at the end of the day, what I wanted to combine two things. One was to make a thriller that people would like to read, and the other was to hit an audience that isn't reading editorials in the Washington Post or the New York Times about some of the things they have to think about. How do they protect their phi their protected health information. How do they keep things out of the medical record that don't want in the medical record. This is not a real what do they call that? Not a teaser? But there is a character in the novel that is a Southern politician that somebody wants to kill. And they figured out how they could kill them through the medical record. I found this fascinating. I didn't know this, but in big hospitals, when you see that ivy bag hanging, you think, well, there's a farm tech or a pharmacist down there who mix it up.

Well, the big hospitals, that's not the case.

They use robots, and the robots are incredibly accurate. But you can imagine a scenario where somebody hacks in the medical record, reprograms a robot and gives me something that is lethal to me. So that's kind of one of the themes that's swirling around in the novel.

Did you find it challenging to learn how to do dialogue and because a novel is really different than writing non fiction, Yeah.

I found it very challenging.

What helped me the most was for several years, I get up early, so I do a little rite in the morning, and inevitably somebody would aggravate me, so I'd come home and write something really nasty about that person. So after about five years, I was so proud of my novel. I showed it to my wife and she said, well, let's see. Now, they're about one hundred characters. There's no dialogue and no plot. Maybe you ought to go off to a writing course. So I did do that. I don't know if you do creative writing or not. It's fun to do because unlike everything that you and I have to do in our lives, there's no background check.

It's just you make it up. So it's fiction.

Where was your writing course?

I went to a course that was put on by company it's not run anymore, called ck Seak. It was on Cape cod and it was hosted by two really great doctor authors, Michael Palmer and Tess Garretts, and both write thrillers. I went to it thinking, well, they're going to make a brief appearance, but they were there for three days. And I also went there thinking, wow, I'm going to get into an issue. This is back in two thousand and seven or eight, because how many doctor writers can they be? And so I was lucky I got in. They said, well, you're lucky you filled it up. They're three hundred and fifty doctors coming, including two classmates mine for medical school, so a lot of interest in this field, and.

Most of them writing fiction or they just want to learn how to write.

Most of them are writing fiction and it ranges from some of the most bizarre stuff you can imagine, to people trying to write things like I was writing.

I've written both fiction and nonfiction in there totally different challenges.

Tell me what you wrote about in fiction, because I love reading fiction.

I started with a very good friend of min named Bill Fortune, who's a history professor but also a professional novelist and has written many, many novels. Our first book was an alternative history in which Hitler is in a car accident just before Pearl Harbor, and so Germany doesn't attack us, and the result is we don't have an excuse to declare war. So we turn to the Pacific, and Hitler dominates in Europe and the novel is called nineteen forty five. And so in this alternity of nineteen forty five, the Germans are dominant in Europe, were dominant in the Pacific, and Hitler has decided he has to attack us, and he has been briefed that there's a facility in Oakridge, Tennessee, developing a nuclear weapon. You've got some fact in there, yeah, And so the whole thing is an adventure story, you know, launching a specialized raid to try to destroy this before the Americans can build it. And I learned so much from Bell because, for example, I would talk with you and say, the Germans had arrived just off South Carolina, and the troops that were going to go to Tennessee had to move from a large ocean going ship to a small ship that was going to take them up river. Well, by the time Bill taught me how you write the last guy trying to get off the ship, and the ship waving back and forth, and the danger of being crushed, and in the end he does get crushed. I mean he took a couple of sentences and turned them into five pages where you're sitting on the edge of your seat, thinking, Oh my god, what's going to happen this. We went on from there and wrote several books about the Revolutionary War from Washington's perspective, and then we wrote a couple of novels about the Civil War, and probably our best gambit, we wrote an alternative history of Gettysburg. And we got the Army War College in Carlisle, Pennsylvania. We got their commanding general and the person who teaches their Gettysburgh class, and we went out and we walked them through our theory because we were both military historians, so we had a theory of how this thing might have happened, and we went out and literally walked the battlefield. And when we got done, the guy who taught the cars said, you know, that actually sounds more like Lee than what he did that day. And so we wrote an alternative history of Gettysburg, which then the two more alternative histories of how the Civil War would have occurred in that setting.

What's that one called?

The first one's called Gettysburg.

Oh, I've seen that book. I've never read it.

Though we did it deliberately because Gettysburgh is the most frequently written about event in American history. We were trying to find some way to launch our fiction efforts. But it was great fun. So when we describe things in the book, that is how they were. You could go and take our book and you could walk the whole battlefield and get a sense for it. But it was really fun, and it was fun to take personalities like Lee or Mead or Longstreet and weave them in in a way that worked well. No, I will tell you I recently did a podcast with one of the great American historians, and his father had written a book called Killer Angels, which is the best book about Gettysburg ever written. It's like poetry. I've written far more of nonfiction than fiction. When you find somebody who has the rhythm and they have the words and they suck you into their world, I find it amazing.

I love it. I'll tell you what.

I didn't know this value, but I'm going to get nineteen forty five in Gettysburg and read him.

When you're looking down the road the next four or five years, do you think you're going to be able to collaborate with the insurance companies and get them to help accept and help implement the kind of changes that are coming with AI.

Well, not knowing who in the world might listen to this, I'm gonna go ahead and tell you what I think, and that is we have tried very hard to collaborate with insurance companies. It's been difficult to collaborate because in so many ways we're at odds financially, and so we haven't tried around AI. I'll tell you one good experience. We had a good experience with Blue Cross, Pushield and Michigan around developing quality initiatives, and this was looking at the quality of different kinds of operations across the state, if you have your prostate taken out, or if you have any kind of cancer surgery. And I think that actually was very productive. We combined our data with their data, their claims data, and it helped identify people that were needed to improve their performance. And it was done in a very protected environment. But I think that was good for healthcare. Subsequently, we tried to collaborate in other areas unsuccessfully, and so AI ought to be one that is un natural. But I'll tell you a funny story I heard about AI recently. So we have one consultant who's telling us that we can come deliver to you AI to help with pre authorization for procedures. So whatever you're going to have, from an eye surgery to heart surgery, your insurance company has to pre authorize that, and that can be a difficult process, and we sometimes end up with somebody who's on the gurney getting ready to have surgery and we still don't have pre authorization, and if we don't get it, we eat the entire procedure in terms of the cost. So I thought, well, this could be really good, And then I was at another meeting of colleagues. They said, man, you got to be careful because the insurance companies now are using AI to figure out how to best deny pre authorization. So, you know, just a great example of AI works in one way for us, another way for the insurance companies. But I think if we can put aside that and think about how we could collaborate in terms of quality outcomes. And it's been talked about for decades, have been the cost curve in healthcare. We've got to do that. We're out of control right now. That's hard for me to say because I've been in healthcare for forty years now. I think AI, in the ways we talked about it before, in terms of better more rapid, earlier diagnosis, better selection of therapies, can make a big difference.

You know, as we think about this much more electronic world. Your book Coded to Kill really centers around the hacking of an electronic health record system. How much as we put more and more things in an electronic format, how much do you worry not just about the criminal hacker, but also the North Koreans of the Chinese. I mean, they are very sophisticated players. IS relatives have showed in Lebanon. I mean, there are people who have capabilities that are a little scary.

How on earth could they do that with the pagers?

You mean it was about a twelve year project?

Really?

Yeah, they ultimately got inside the supply chain and made sure that only their got to Hesbela. It's a very elegant, very difficult project, and it takes a country that's in desperate efforts to survive. But if you think about it, all these different kinds of things, how do you build layers of defense so you can't be hacked into?

Right?

Well, I'm happy to both tell you and here you critique it. Given your knowledge of national security, as you know pretty much what we're thinking about. So when I came here, one of the first questions I asked was how often do we have people inappropriately accessing our medical records? When I was at UNC, I got into this because the dean at the time we had a bunch of people mini staff and about forty faculty who had inappropriately accessed the electronic medical record of one of the fabulous basketball players who was out for a few games. And so, lucky me, I got a sign to talk to all those faculty about this and you know how it's a federal crime, and try to scare the dickens out of him.

You know, I got the wildest stories.

One guy told me, well, sickle cell experts, so I worried he might have sickle cell disease. So I was looking to see if I could be helpful. I mean, that wasn't it. He's a sports fan. I got here and I said, does that ever happened at Michigan? They said, oh, no, no, no, we know it doesn't. And I said, well, what automated system are using to look at every medical record? There are some out there, and they weren't using one. It took two or three years to get them to use one, and turned out we were having many, many people that were inappropriately accessing medical records. Now that's not a hack at all. They just have access. They find out you're in the hospital, or the governors in the hospital, they say, huh, let me see what's going on. And so we fired a whole slew of people. It's taken about five years to get it under control. Now we're down to about twenty times a year that that happens. But that's trivial. What I worry about is we get about between six thousand and more attempted hacks a day externally, and most of those I think of as like those robo calls you get during dinner time, but they're about eighty a month that are serious attempts from foreign actors and sometimes domestic. And what's said in healthcare, which I believe is sadly true, is it's not if you will ever get hacked, it's when you'll get hacked and what you do. And so the layers we've tried to build, and we have lots of security on top to try to permit the hacks. But some of the best things we're doing, and is AI enabled, actually is to detect inappropriate access into our medical record and immediately coordinate off. So, you know, the CrowdStrike thing that happened recently, we were back within hours, a couple of hours, where somehow systems in Michigan were down for two or three weeks. You know, we can't do anything when your EHR is blocked. And our university had a big hacked about a year ago during student registration time that was in the medical school, and as the university, it's like a pimple. How do we surround it? So it doesn't spread. That's our strategy. I'd love to hear what you think we ought to be doing.

The only advice I would have is to find one of the most recent national security agency heads and bring them in and ask them to review everything we have not done.

That. That's a great idea.

I'm very close to one of the former in and SAY heads. That's a world that is so sophisticated, and it's still the leading capacity and an internet activity in the world. Marshall, I want to thank you for joining me. This has been a wide ranging and I think pretty fascinating conversation. Your book Coded to Kill, I suspect is the beginning of.

A whole new career, I hope.

So that's available now on Amazon and bookstores everywhere, and I really appreciate you talking with us. I hope we can do this again when your next book comes out.

Well, thank you again, it was awesome.

Thank you to my guest doctor Marshall Rangy. You can get a link to buy his book Coded to Kill on our show page at newtsworld dot com. Newtwell is produced by Gager three sixty and iHeartMedia. Our executive producer is Guernsey Sloan researcher is Rachel Peterson. The artwork for the show was created by Steve Penley. Special thanks to the team at gingrid Street sixty. If you've been enjoying Newtsworld, I hope you'll go to Apple Podcasts and both rate us with five stars and give us a review so others can learn what it's all about. Right now listeners of Newsworld consigner for my three freeweekly columns at gingridsweet sixty dot com slash newsletter. I'm Newt Gingrich. This is Newsworld.

In 1 playlist(s)

  1. Newt's World

    822 clip(s)

Newt's World

Join former House Speaker, professor, historian, and futurist Newt Gingrich as he shares his lifetim 
Social links
Follow podcast
Recent clips
Browse 815 clip(s)