The Rich Roll Podcast - The Dean of Stanford Medical School on How AI Is Shaping The Future of Health Precision

Episode Date: April 11, 2024

This week, I am joined by Dr. Lloyd Minor, the Carl and Elizabeth Naumann Dean of the Stanford University School of Medicine and Vice President for Medical Affairs at Stanford University. Dean Minor talks about the transformative potential of AI in healthcare delivery, research, and diagnostics. We discuss its nuanced pros and cons, including impacts on accessibility, safety, and efficiency. Dean Minor examines AI's benefits in drug discovery, Precision Health, and early disease detection. He elaborates on wearables and the shift towards a proactive approach, integrating tools like virtual reality into medical education and emphasizing nutrition in training. Addressing ethical considerations and industry influence, we delve into the regulatory framework driving transformative changes. We also explore groundbreaking diagnostics, envisioning a future revolutionized by growing and 3D printing organs, and much more.  Enjoy! Show notes + MORE Watch on YouTube Newsletter Sign-Up  Today’s Sponsors: Brain.fm: Focus music for productivity—listeners can get 30 days FREE  👉brain.fm/richroll Waking Up: Get a FREE month of mindfulness resources plus $30 OFF  👉wakingup.com/RICHROLL  AG1: Get a FREE 1-year supply of Vitamin D3+K2 AND 5 free AG1 Travel Packs 👉drinkAG1.com/richroll Faherty: 20% OFF your first order when you use the promo code RR20 👉FahertyBrand.com/RICHROLL  Roka: Unlock 20% OFF your order with code RICHROLL 👉ROKA.com/RICHROLL Go Brewing: Use code Rich Roll for 15% OFF my favorite non-alcoholic brews 👉gobrewing.com/discount/richroll

We ought to be able to get a better picture moment to moment on the status of our health and therefore be able to act upon early signs in ways we haven't been able to do in the past. How is technology, specifically artificial intelligence, changing the landscape of medicine and healthcare? There will be some revolutionary changes in diagnostics, revolutions in drug discovery, better therapeutics, and that's going to improve health and well-being. Dr. Lloyd Minor is the Dean of Stanford Medical School and also serves as Stanford University's Vice President for Medical Affairs. I do think

medical education is going to be significantly different a decade from now because of large language models and just the information that they bring to people's fingertips. He believes AI is medicine's biggest moment since the invention of antibiotics. There's never been a better time to be in the life sciences than today. We're training a model based on a lot more data than an individual pathologist will be able to see in their lifetime. We also discuss, of course, the ethical considerations this technology demands, as well as the many dizzying ways AI isn't just changing the healthcare game, it's actually creating a new game altogether.

It's my privilege to share his insights with you today. So here we go. This is me and Dr. Lloyd Miner. Thank you for doing this, Lloyd. I appreciate it. Thank you, Rich. I'm honored to be here. Really have enjoyed following your podcasts and learning from you and your guests.

I appreciate that. That's very meaningful to hear. So thank you for that. I think before we get into the subject matters at hand, I'd love to kind of better understand your job. What does it mean to be the dean of Stanford Medical School? Like elaborate on what that role is and what your responsibilities are. Sure. Well, fundamentally, my job is about people, and I have the privilege of working with some truly amazing people every day. People, faculty. Faculty really do everything that's part of our mission.

And what is our mission? It really has three parts, patient care, research, and teaching. And those three components are synergistic. So the research we do drives advances in patient care. Students, medical students, PhD students, master's degree students, clinical fellows and residents, they come to learn from our faculty and to be able to advance their skills and their knowledge and also learn the art and the science of medicine for those that are involved in clinical medicine. So my job is, broadly speaking, to work with the people that make this happen on a daily

basis, to make sure that we're garnering the resources that enable the people to succeed, and also to help in setting the strategy for the overall enterprise to lead the strategic planning initiatives that are part of the enterprise. I would imagine it's a delicate balance to kind of attend to the specific interests of these various pillars, the business of the medical school, the clinical aspect of it, the patient care aspect of it, because those interests don't necessarily always align. And so it centers you in sort of a political role where you have to kind of navigate the various interests and kind of do that amidst a group of people who are diverse in their interests,

aims, and goals. That's right. That's why the job is never boring, which is wonderful. You seem to have a very, you know, kind of genial disposition, though. I think you have to be. I think also one has to be very positive. There's a lot of reason to be positive. We have to look at challenges as opportunities. And I try to understand, and yes, there are situations where interests are not aligned. And that, just as you point out, is somewhat inherent in the nature of the enterprise. There are so many related but sometimes divergent goals and initiatives within the umbrella of an academic medical center and a school of medicine.

And the job of leadership is to work as hard as we can to harmonize those. And in most cases, you can find synergies when you look hard enough and you look beyond the apparent conflicts. But there are conflicts, and there, you try to approach them with an attitude of fairness and looking for ways in which everyone succeeds, even if not everyone gets exactly what they came in desiring to have. Well, that's a keystone of negotiation, right? It is. Having everybody walk away with some kind of win, right? Precisely. Yeah.

Precisely. Yeah. Precisely. Well, you are here because you are at the forefront of this brand new technology that we're all kind of inelegantly trying to wrap our heads and minds around, which is artificial intelligence. So explain to me how this first kind of became interesting to you and what convinced you that this is the massive breakthrough that you present it to be in terms of medicine, medical education, diagnostics, health care, basically everything. Yes. And, Rich, I think you summarized it well. It does encompass so many things. And the use of artificial intelligence or the elements of artificial intelligence in the delivery of health care dates back quite a while.

For example, today, when we go to a physician or other health care provider and we get a prescription for a medication, it would be very rare today that we actually get a handwritten prescription that's mostly entered, and certainly in the hospital setting, always entered electronically. What moving to electronic patient records and electronic ordering systems did was to enable us to be safer in how those prescriptions are written, delivered, filled. So now an error, which can happen, a decimal point gets moved. Sometimes maybe someone prescribing a medicine doesn't know, doesn't realize that the person's on a medicine that actually has an adverse reaction to the medicine they're prescribing. Now all of that gets reconciled electronically. So that's a form of crude, rudimentary form of the application of artificial intelligence

to a process, namely prescribing of medications. As our ability, as the algorithms driving AI have evolved, and as the systems for taking in vast amounts of data have grown, now we're able to apply AI in ways that we couldn't have even dreamed about five years ago. The more traditional forms of AI that usually are described with machine learning or deep neural networks. And now more recently, there have been the large language models or transformer models that I think open a whole new vista of opportunities in healthcare, fundamental life sciences discovery, and pretty much everything that underlies our business. So what are the pros? Let's get into the pros before we talk about the cons and the ethical dilemmas that are presented? Like, what are we looking at in terms

of what we can imagine with this power now at our behest? I think the pros are that AI will help healthcare delivery to be more accessible, to be safer, that AI will help discovery all the way from the basic science, fundamental discovery that's done in labs, all the way up through the design of clinical trials, that it will make those processes more efficient and more effective. And maybe if I could, I'll describe sort of each bucket separately. So on the healthcare delivery side, making it more accessible and equitable and safer, a few years ago, our dermatologists worked with some computer scientists. And what they did was to take pictures, just with the smartphone, of skin lesions,

and then annotate those photos with what we knew to be the pathology of the skin lesion. Like, was it a cancer or was it not? And then they trained a neural network from a group of many, many pictures. And then they presented to that neural network, that AI model, new pictures that hadn't been used to train the model of skin lesions and ask, is it cancer or is it not? And they did that with the AI model. And then they asked a group of, you know, board-certified dermatologists to look at the same pictures. The AI model was as good as the board-certified dermatologists from looking at pictures and discerning whether or not this was a malignant lesion or not. Now, what does that mean?

It means that, for example, in a rural area that may not have a dermatologist, or even in an urban area, it's not always easy to get an appointment to see a dermatologist. A primary care physician can, with a picture of the skin lesion, get a pretty good indication about, oh, this is something serious, and I've got to get this patient in to see a dermatologist, or this is probably nothing to worry about. We'll just watch it. So that makes healthcare more accessible, and ultimately, I think, should make it more equitable. I can feel the sort of skin on the back of my neck prickling a little bit because we've all had that experience of toying around with the LLMs out there. And on first glance, you get an amazing result. You're

very excited. And you kind of dig a little bit further. And very quickly, you realize like, oh, this might not be as advanced as I originally thought. Errors abound, et cetera. And I'm imagining a situation where the photograph isn't taken quite right or the light isn't exactly correct. You get a misdiagnosis. So to your point around safety, those are the alarm bells that go off in my mind. But they're the same alarm bells that go off when I think about the prospect of autonomous driving or any of these other technologies that are on the horizon that sort of threaten our illusion of control and safety on some level. Exactly. And all good points. That's why we have to roll out this technology with a lot of oversight, with a lot of insight,

and very importantly, with the engagement of the public. This can't be sort of top-down. We have to be very transparent with the public about how AI is being used. And we have to get feedback on that as it is rolled out. But let me mention another example. Lots of diagnostic imaging studies are done every day across the country. Now, a diagnostic imaging study, whether or not it's a chest x-ray or a CT or an MRI, Now, a diagnostic imaging study, whether or not it's a chest X-ray or a CT or an MRI, all that data comes in digitally. We stopped printing X-ray film years ago. Since it comes in digitally, it can readily be used for AI.

across the board of diagnostic imaging, roughly 4% of the interpretations of those images by humans miss something that's of clinical significance. That doesn't always mean that the patient's going to have an adverse outcome, but something is missed. Now, if AI, as it's being used today, for example, in chest x-rays, if it can be used to trigger, you know, say to a radiologist, look at the upper lobe of the lung because there appears to be something there. If it can also look back through all the old chest x-rays and say, well, this was there three years ago and this is how it's changed, it's helping. It doesn't supplant the radiologist from making the final decision, but it helps to prevent an error and make the delivery more efficient. To your point around the efficacy of the AI model versus the standard practitioner or specialist's ability to kind of detect based on experience,

I would imagine there might be a little pushback from those specialists who are adamant that they're better at doing this than the model. In the same way, to extend the autonomous driving metaphor, we think we're better at driving than the robot version, but all the data and statistics suggest otherwise. Is that something that you have to contend with as you go out into the world and talk about these things, that it's the doctors themselves who might be bristling? Definitely. And we've seen this before in healthcare and medicine. For example, when I was in my clinical training after medical school, this was in the early days of angioplasty. And at the time, cardiac surgeons thought angioplasty is never going to go anywhere.

You know, these are all going to fail. And these stents, they're going to get clogged up. Well, angioplasty and cardiac stents have saved many, many lives and continues to get better. And, yes, there is still an important role for cardiac surgery. Cardiac surgeons haven't gone away. Likewise, in general surgery, it used to, when we took out a gallbladder, we made a big incision and took out the gallbladder. Now many of those procedures can be done laparoscopically with a tiny incision and much shorter recovery time.

What happens over time is that practitioners retrain and medical specialties redefine themselves. I don't think radiologists are going away, not at all. But there'll be radiologists who use AI in a responsible way and those that don't. And in the end, the ones that do use it will be the ones that are in practice. The immediate and kind of most obvious use case that I see being such a huge benefit is the ability of these models to drugs for specific purposes or to design or refine clinical trials so that they're better suited to the goals of the scientists who are conducting them. Exactly. And, you know, we talked about the use of AI in radiology, and we talked about that proceeding at a rapid pace because all the data comes in digitally.

But I think the impact perhaps is even going to be greater in pathology. So when a tumor is removed, when a growth is removed, it goes to the pathology lab and it gets sectioned, and then a pathologist looks at sometimes even hundreds of slides from that tumor. sometimes even hundreds of slides from that tumor. For rare conditions, a trained pathologist may have only seen a half dozen, a dozen of a particular lesion or abnormality in their career. And what you were just saying, because now we can collect from a variety of different health systems hundreds of these tumors that are rare, health systems, hundreds of these tumors that are rare, we're training a model based on a lot more data than an individual pathologist will be able to see in their lifetime. So that increases accuracy. It increases precision in knowing what the salient features of a tumor are in ways that no one human being or group of humans can do.

What are some of the other benefits? I think you mentioned drug discovery, for example. One of the early applications of machine learning was in the study of protein structure. And now we can predict the structure of a protein just by knowing its sequence very accurately. That's enhancing the drug discovery process. We're already seeing benefits of that. We also have the possibility in the future of designing a drug, if you will, from scratch, just from data about the biology of the condition that's being treated.

And there are companies that are focused on that today. biology of the condition that's being treated. And there are companies that are focused on that today. So I do think we will see a revolution in drug discovery. How quickly will that come? That remains to be seen. But the pieces are there to have that sort of an impact. And the better these models get, the more refined and improved their diagnostics become, which translates into earlier detection, right, for these diseases. So the real world kind of ramifications are there's an expedited time period from diagnosis to treatment, but also the early detection piece, which obviously is going to, you know, help people resolve these problems before they get too far out of hand.

is going to help people resolve these problems before they get too far out of hand. That's exactly right. It's back to a concept that we've talked about before in Stanford Medicine and elsewhere, a concept we call precision health, which we distinguish from precision medicine. Precision medicine is about getting the right treatment to the right patient for the right disease at the right time. But precision medicine is, if you will, after the fact. It's after someone gets sick. How do we provide the best sick care? And of course, in the United States, we have the best sick care system in the world because of our specialized care that we offer because of tertiary and quaternary care.

care that we offer because of tertiary and quaternary care. But we haven't focused nearly as much attention on predicting and preventing disease and in detecting it earlier, as you just mentioned. And I think these applications of AI to bettering diagnostics and designing more predictive diagnostic tests. You know, every time we fly on a plane, the engines of that plane are being monitored hundreds of times a minute on the ground. We should be able to do something comparable to that. One of my late colleagues at Stanford, Dr. Sam Gambier, very much had that vision. We should be able to do something comparable to that in diagnostics and develop something along those lines. Well, we're slowly inching towards that with the advent of tech consumer technology

in the form of like wearables, I've got a whoop on, I use inside tracker and we've worked with levels and I'm going to experiment with super sapiens these cgm companies uh apple health with the apple watch and everything that they're doing and i know you had this this heart study with apple i want to hear about that but i'm curious around the future applications of how the data sets that are extracted from all of these wearables can be used to better predict and better prevent. I'm thinking I had Tim Spector in here who founded Zoe, and he was talking about, you know, the massive amount of data that they're able to kind of mine through the Zoe app,

which allows them to come up with new therapies, predict diseases, et cetera, and create, you know, to come up with new therapies, predict diseases, et cetera, and create diagnostic tools as a result of that. And I think with mass adoption, you're only going to see more and more and more of that. It brings up privacy concerns, of course, but where do you see all of that going? I think wearables have an important place,

and you're right, we did work with Apple several years ago on the Apple Heart Study. The question we addressed with the Apple Heart Study was, can the Apple Watch be used to detect the most common heart arrhythmia, atrial fibrillation? And this was a very large study, several hundred thousand people participating in it all virtually, and very rigorously controlled. And yes, the Apple Watch can be used to detect AFib. That's useful because in many cases, people don't know. Some people have an indication, shortness of breath or other indication when they go into AFib, but in many cases, they don't. And we know that people that stay in AFib for any prolonged period of time are at a higher risk of stroke and various other conditions. We know that people that stay in a-fib for any prolonged period of

time are at a higher risk of stroke and various other conditions. Now that is relatively low-hanging fruit in terms of how... It's rudimentary. That's exactly right. But that opens the door. We do have glucose monitors available today. I think those will become in more common use. There's ongoing work in monitoring blood pressure in a non-invasive, real-time way. We ought to be able to get a better picture moment to moment on the status of our health and therefore be able to act upon early signs in ways we haven't been able to do in the past. I think the real barrier to true mass adoption

is the fact that right now, the marketplace is so diffuse and dispersed. And it's like, how many of these things do I have to put on me? I've got this here and that there. It's gotta be integrated. And if history tells us anything, Apple's pretty good at that.

They sort of take this wait and see approach and then they suddenly strike, you know, when the iron is hot and when the time is right and kind of take over. I could imagine a scenario like that, but I really do think that that is the future. And I think that these devices, if they've taught me anything, it's given me a sense of agency, like the transparency and understanding what's happening with my body in real time is information that is empowering. It's a little scary. Like, do I really want to see what's going on with my insulin levels right now? And certainly there are people out there that kind of push back on consumer CGM adoption. But I think it's just, I think that that can be overcome with education so people

understand what it really means when you're looking at this stuff. And I can't imagine a future where this isn't going to become more and more integrated into our lives. I agree with you entirely. You know, I think the other thing that is going on related to what you're saying is that in the past, in the United States in particular, we've had, as individuals, we've had a passive attitude about our health. We've always assumed, well, if we get sick, we'll go to the doctor and we'll get a medicine and we'll be fine, right? And I have to say that we in healthcare have, in the past, we've, in some cases, we've encouraged that attitude. Really, what we want is for each of us as individuals to first and foremost be responsible

for and engaged in our health. And the role of us as physicians is to partner with our patients, yes, to have a base of knowledge and skills that help our patients, but fundamentally to enable our patients to take care of themselves. And furnishing information that's actionable to us as individuals is an essential part of being able to take care of ourselves. And then helping people understand what that information means. So does that transient spike in your glucose level when you eat a donut, is that going to have a long-term effect if, say, your hemoglobin A1c is normal?

Is that going to have a long-term effect on your health and well-being? Well, probably not. And by the way, if you just focus on keeping the line straight, you may be- That's gonna drive some unhealthy, if you're just eating saturated fat all day, because it keeps that, if you gamify it, you can end up in a not so good place.

But I think to your point, when you look at what's really killing most people, it's these chronic lifestyle ailments, it's heart disease, it's these chronic lifestyle ailments. It's heart disease. It's type 2 diabetes, obesity, the increasing rates of Alzheimer's and various forms of dementia that seem to be kind of metastasizing right now. And understanding that these illnesses don't happen overnight. They are growing and taking place over a period of decades. And so if you can detect that something's not right 20 years before the heart attack, you're in a pretty good place to change directions and avert that disaster that is the thing that kills most people.

Exactly right. And of course, that's what- But our healthcare system isn't really set up for that right now. It's not. It's not. I mean, the example you draw in terms of being able to prevent a heart attack by having your cholesterol level checked and then doing things to lower it, that's a good example of proactive preventative care that has been effective. And the instance of heart disease has been steadily declining, and our ability to treat it has been increasing. But we need to generalize this across the board, and we all need to be more engaged in healthy lifestyle and evidence-based interventions ourselves in maintaining our health.

Talk a little bit about how these technologies are reshaping medical education. What is the experience of the typical med student now versus where you see it going in five years, ten years as a result of this tech? Well, I'll give you an example. When I was in medical school, not only did we memorize the names of drugs and their mechanisms of action, we had to memorize the dosages of drugs. That was crazy because the dosage is an arbitrary number. It depends upon how the drug is formulated and everything. You can't keep arbitrary numbers in the human brain. But nonetheless, that's what we did and what we were tested on. Well now, and I talked about electronic prescribing before, we certainly don't memorize

the dosage of drugs. We still need to know mechanisms of action and how drugs maybe interact. But in the future, the need for memorization and the need for an active working knowledge base, I think, is going to diminish. What we're going to have the need for is really understanding how to use the data sources that are out there, how to be skeptical when those data sources aren't giving us. You mentioned the hairs on the back of your neck raise when you hear about how AI is being applied to interpreted images, we need to make sure that physicians have a lot of skepticism about the information that they're being given from AI, but we also need to train them on how to use it responsibly.

So back to your question about medical education, I think we'll continue to see a de-emphasis on memorization, because still there is a lot of memorization in medical school. emphasis on memorization, because still there is a lot of memorization in medical school, will, in the discovery aspects of medical education, will be training students to use these AI models to ask questions and get informed answers to those questions, and then to drive either in patient care or the research they're doing based upon their interaction with AI models. But I do think medical education is going to be significantly different a decade from now than it has been in the past because of large language models and just the information that they bring to people's fingertips.

Has virtual reality found its way into the medical school curriculum? Because there's so many use cases you can imagine of putting those goggles on and participating or observing a surgical procedure or actually going inside the body. Is that happening already or what does that look like? Absolutely. Let me mention two examples, and these are commercially available products. We at Stanford still have our medical students do cadaver dissections. We feel like that's an important part of learning. It's also an important part of developing the culture of respect for the human body. But those dissections are supplemented now by virtual reality and increasingly augmented reality approaches to really understanding the three-dimensional aspects of anatomy.

So there are electronic simulations where you can look at any plane you want to in the body. You can insert specific muscles, take away muscles. You can understand the pulling direction of muscles much better than you can from a cadaver dissection, certainly much better than you can from a textbook. So that's being used today. And in a more direct clinical application, our neurosurgeons have a system that, I mean, operating the brain is complex.

And every tumor, every structural abnormality is a little bit different. So the surgeon can work with a virtual reality set to actually do the operation virtually, you know, based upon where the tumor's located, and be able to see the relationship of the tumor, for example, to blood vessels and things like that before they ever get into the operating room and have a much better understanding, okay, these are the steps I'm going to need to take. This is what I'm going to need to watch out for at this particular point of the operation. So it's almost like a flight simulator version of a surgical procedure. That's right. Where you're having an almost tactile experience

of doing it without the risk. Exactly. Yeah. That's right. Where you're having an almost tactile experience of doing it without the risk. Exactly. Yeah, that's amazing. I gotta ask, what is the extent of nutrition education in medical school? I've had so many doctors come on this podcast, we had one elective class of nutrition education in medical school. I've had so many doctors come on this podcast. We had one elective class of nutrition, or we only had to do four hours or something like that over three years.

Is that changing? What does that look like? I mean, I know Stanford Medical School is very progressive in this regard. I imagine this has been considered, but it seems like more broadly, it still remains in a bygone era. It is increasing. It needs to increase more. I know you've had Dr. Christopher Gardner as a

guest on your podcast, and Christopher is amazing and has been a real champion for introducing nutrition education into our medical curriculum and more broadly. I think it gets introduced in several contexts, but I'm not for a moment saying that we shouldn't be more focused on it. But it is being introduced now in the standard ways that we teach carbohydrate metabolism, for example. But I think integrating it as well into the clinical curriculum to know how to talk to patients about their nutrition. And, you know, I met a faculty member who a number of years ago did a study looking at obesity to see if offering to a family to replace the cooking utensils in the home with smaller utensils, reasoning that if they're smaller utensils and you cook less, you'll eat less,

if that might lead to weight loss. And lo and behold, yes, it does. So we need to be thinking at all levels of the medical education and care delivery process, how we build in a focus on nutrition, both scientifically and educationally. Healthcare is sick care, to your point. And there's certainly a lot to redress when it comes to our health care system. I want to put a pin in that for now, but I bring it up because if we want to truly move towards this new modality of predict, prevent, we have to instill that in the medical education, right? So prevention and prediction can be done with all of

these new tools better. Prevention often has to do with all these lifestyle interventions, right? Which includes nutrition, but has other things. And so the common refrain when someone goes to the doctor is that they never make any kind of lifestyle recommendations. It's not really part of what they do. They're time constricted. It's not necessarily their fault. It's a systemic thing. But the more that we can educate our young fledgling doctors around these things, the better chance we're going to have that they're going to carry these principles into their practice and share that with their patients so that we can be more in a preventative stance when it comes to healthcare outcomes. Exactly. And when we talk about prevention, we have to first and

foremost look more carefully at behavior because prevention is fundamentally about changing behavior. I'll tell you a story. About a decade ago, shortly after I moved to Stanford, to California, I was getting to know people in our community. I met with a leader in the life sciences venture community. And I said, is there any topic that if someone comes in and tries to pitch you on a company doing this, that you just say, you know, thank you very much, appreciate it, but I'm not interested.

And he said, yeah, anyone who comes in and tries to pitch me on something that's going to change behavior, I'm not interested. And he said, yeah, anyone who comes in and tries to pitch me on something that's going to change behavior, I'm not interested. I was like, oh, dear. Because now we've seen that change, right? We've seen some real successes in the digital health world. job of integrating the study of behavior, the study of lifestyle, study of well-being, much more into the scientific mainstream of research, and also into our attitude and our approach to care delivery. Because behavior really does underlie a lot of, we mentioned chronic diseases before, that are so crippling in our country right now.

Every one of those will have a strong behavioral component. When it comes to more progressive modalities around medicine, you hear about functional medicine and integrative medicine, preventive medicine, holistic medicine. Now we have precision medicine. How do we parse all of that? How is what you're talking about with precision medicine different than those other terms? There are many, many similarities, but in particular by focusing on health. Our goals with precision health is to use the same enablers that have been used for years in precision medicine. You know, let me mention an example.

We are so much better today treating breast cancer than we were a decade ago. Why? Because not every patient with breast cancer gets the same treatment. The treatment is tailored to the tumor and to the individual. And as a result, we have much better outcomes. to the tumor and to the individual. And as a result, we have much better outcomes. What we should be doing with precision health is taking that same knowledge base, knowledge base of genomics, of lifestyle and other things, and applying that in a predictive way to say,

well, maybe I need to have these screening tests done every year. Maybe another person needs a completely different set. Now, we already do that to an extent, but we should be doing it to a much greater extent. Underpinning a lot of this will be advances in diagnostics, which I think are coming along, but diagnostics are hard. First, you're trying to detect a very small signal, and you need to do it accurately because you don't want to drive a lot of unnecessary testing because of a false positive. But that work's being done, and I think it is going to lead to more tests that have the same sort of actionable value as when you and I have our cholesterol measured, for example. So those are going to be the things that we need to

really drive this revolution in prediction and early detection. Right now, medicine is divided up amongst all these various specialties, and it's unclear how much sharing of information or cross-communication there is between all of these fiefdoms. And one real advantage that I can imagine with these AI tools is that they can take massive data sets from genomic testing and sort of cross-section that with microbiome data and metabolic health data from CGMs, like across the spectrum of all of these different specialties, and try to make sense of how they fit together and how that can drive better predictions and better early diagnoses. That's right. And that's another strong reason why AI is going to be transformative, because bringing together those different sources of data

is generally more than the human brain or even any groups of humans can do. Yeah, we're not capable of doing that. No, exactly. Those data exist, and collating them through the benefits of AI and deriving information from the troves of data that exist out there, particularly in electronic health records, I think is a major goal of the application of AI to healthcare delivery. Let's talk a little bit about the cons or the ethical dilemmas that come up as we move, you know,

towards this near future and the considerations that are underway to kind of address them. I know you have partnered, you have this organization, it's called RAISE Health. That's right. So talk a little bit about that and, you know, where your head is at. Well, RAISE Health is an acronym for Responsible AI for Safe and Equitable Health. And I think that says it all. How do we deploy AI in an equitable way and in a safe way to improve the health of all of us? So some of the cons, some of the risks that we need to be clearly aware of and mitigate and prevent. First is we have to protect privacy. And there are new dilemmas that are going to arise in privacy that we haven't had to deal with in the past For example, with the large language models that exist today And particularly as those language models start to integrate social media data

You might have a doctor in an emergency room at 3 o'clock in the morning Seeing a patient who just returned from a trip to South America who has lupus and a few other medical conditions, and this patient has a high fever. And you type in that information into a large language model, even though there's nothing in that that's identifiable. There's no personalized health information in that query. And still, by linking various sources of data, the patient could be identified just from that query. So we have to think about privacy in a new way. And there are many, many ways of doing it. One is to bring the model into an individual delivery system so that data doesn't get out.

You can ask a query and it doesn't get back to a model that's going to be trained based upon the data. Right. That's the whole, like, don't worry, we're keeping the AI in the box thing. You know, that is the premise of every dystopic, you know, super intelligence movie you've ever seen. True. But that's something, privacy has to be front and center. And in President Biden's executive order related to AI that came out just recently, privacy was at the top of the list in terms of what the regulatory agencies have to be focused on protecting.

So that, to me, is at the top of the list for the downsides or the things we need to protect as we roll out AI. There are others as well. The AI is only going to be as good and as accurate and as reliable as the data used to train it. And if the data used to train AI is biased because it only contains data from white men, for example, then it's going to yield results that conceivably could lead to inequitable recommendations. So being cognizant of the risks of bias and mitigating those risks.

Of course, mitigating those risks means doing studies that are inclusive, more inclusive of populations that historically have not been included in medical studies or clinical trials. So bias is another example. I think the other thing we want to protect is the primacy of the relationship between a patient and their health care providers, their nurses, their physician assistants, their physicians. care providers, their nurses, their physician assistants, their physicians. If AI supplants that or is somehow viewed as a gatekeeper to getting to a human being, then that's not a good thing. I don't think it will do that.

I think it actually will make the interaction between a health care provider and the patient more connected. We probably all had the experience, I certainly have, of going in to see a healthcare provider and immediately they're typing the note. And even though, yes, they're listening, right? That's not the same as the conversation you and I are having across the table from each other looking at each other in the eye. Right. If you can remove the clipboard and make better eye contact and be present with that patient, that's its own healing modality. That's right. And that's within our grasp. So those are some of the things that

I think are the downsides that, again, a goal of Raise Health is not to try to say that those are not a problem, but indeed to say they are an issue, raise them to the fore and responsibly address them. Another goal of Raise Health is to make sure that every step of the way we're communicating with and engaging the public. There was a study done last year by the Pew Foundation looking at, asked the question, a general question of a large number of individuals, do you trust the application of AI in the delivery of your healthcare? Not surprisingly, overwhelming majority of people said no. Of course, they should say no, because we haven't explained to them how it can be used, how it is being used today, for example, to prevent medication errors and

other things, but also to understand what it offers in the future and how we have to work together to prevent the downsides. It's sort of like there's no pilot in the cockpit. You know, I'm imagining a situation in which robotics and automation are in the hospital ICU room and they're calibrating what ends up in the IV drip, for example, or they're administering drugs without any kind of human involvement. And the potential for that to go haywire is scary enough to kind of make anyone fear that type of future. For sure, for sure. But I would mention, though, that, again, to your example, the use of robotics in making up a drug formulation,

that's a good example because chemotherapy infusions typically have to be very accurately calculated to someone's weight, their blood count. When we were relying just upon humans to do that, there definitely was more room for error. And not only room for it, there were more errors. Hmm. So interesting. How are the other big tech players in the private sector considering all of this? I'm thinking about Walmart and Amazon, like Amazon is ripe to sort of get into healthcare in a really big way. Walmart with its, I don't know how many stores they have, but as a vehicle for

the greater access that you're talking about, it seems like they could be, you know, sort of critical partners in how we revolutionize access. I agree. Walmart, for example, has at many of its super stores, they've chosen markets to roll this out in where they can study it, and also based upon sort of their assessment of where it will offer benefit. But they have clinics next to the super store. And they have been very, as is the case in other areas, they've been very transparent about pricing. How that will generalize to other, given Walmart's closeness to everyone in America in terms of having stores, that remains to be seen.

Amazon acquired a primary care delivery system in its recent history. Now, will they generalize that? Amazon, I believe, is getting into also filling prescriptions. Walmart has a very large pharmacy service. So I think the large retailers are definitely looking at ways to make healthcare delivery more efficient and more accessible by placing it in communities. The larger tech firms, I think, are looking at where they can have impact. We talked about the Apple Watch study before. We, Stanford, worked with Duke and with a branch of Google or Alphabet called Verily to roll out a project that we call Baseline. This is a study of several hundred individuals where everything about their health has been measured and they're being tracked.

There's a wearable that was developed specifically for this study because we want to be able to look at a lot of data coming in and to be able to say, well, there was an early indication three years ago that we didn't even realize was an indication that maybe was associated with a condition that the person developed. And the only way you can develop those relationships, discern those relationships, is by studying things what is called longitudinally, over the course of several years. You know, there are studies like that going on. You know, a few things about healthcare that's a challenge for tech. One is that it's by far the most highly regulated industry in America. And that makes doing broad innovation and application and rollout a challenge.

So I think tech firms are being careful of that. But there's no question, I think, that there will be a bigger role for technology. The wearables we were talking about before in the future, it just has to be done. It's not the same thing as rolling out a new version of a smartphone, for example. Sure, sure. In the same way that we all kind of woke up one day and were informed or realized that everything that we had been doing on social media, chatting with our friends and scrolling and uploading photos and replying and commenting, et cetera, was not only being tracked, but being mined to such a fine degree and then repurposed and sold to third parties so that we could be

advertised to was a disturbing revelation for most unsuspecting people. But what you're talking about is the next step of that. Not only are we going to monitor everything you do on social media, we're going to monitor your blood work, your heart rate variability, your sleep cycle, your insulin levels, and everything in between. And that data will be used for your own good because we're going to do good with it. That data will be used for your own good because we're going to do good with it. It's not hard to imagine why that could be a difficult sell for a lot of people. For sure. Right now, where the pop-up comes, do you accept these cookies? I don't know about you, but I typically just say, I want to get on to whatever I'm doing.

Yes, I accept the cookies. We're going to all have to be much more careful about how we allow or don't allow our health data to be used. I mean, much more careful compared to just clicking, yes, we'll accept cookies or no, we won't. That gets back to this privacy imperative that we have to have. And it does mean that the impact of these sorts of data mining technologies on the actual delivery of care is going to proceed more gradually for good reason than, for example, when I go on Amazon and search for shoes, then all of a sudden when I'm on the New York Times website, I have these ads for shoes popping up. I mean, for sure you can't have that.

But now a breach of that trust would be, hey, this guy's not sleeping so well. So suddenly you're getting ads served up on your social media accounts that are advertising supplements to improve your sleep, et cetera, ad infinitum. Like this is not a world that any of us want to be in. So safeguarding that data and perhaps even creating a situation where somebody has to opt into that as opposed to opt out. Imagine wearable company X, they're sitting on all this data. We can just sell all this data for a lot of money because there's people out there that can use this to market and pinpoint these people to sell them things. Exactly. And that has to not occur. I mean, we have to protect it. But, you know, Rich, the history of, and I know you're not suggesting this, none of us is suggesting that we just close the box on applications.

That's not happening. That never happens. It would be the first time in human history that we've closed the box. So I think we can all accept that the box is not getting closed. Right. And it's just going to be what it's going to be, and we're going to try to put the guardrails up as best we can. But history also tells us that human hubris always believes that we can better

control things that we find out later that we didn't do such a good job with. So how is this going to be different, Lloyd? And I'm pushing back in sort of in fun jest, but I am curious, and I think these are really important issues. And I know that you're considering them deeply, but these are the things that people are going to want to know if you're expecting them to feel safe and to, you know, toggle that opt-in button. Exactly. And one way it's going to be different is because you and I are having this conversation today. I don't think with a lot of technology innovations that have impacted our lives, the conversation is generally post hoc. It's after the fact. Oh my goodness, look at the effect that social media has had on, and then fill in the blank, our children. And I'm not taking a side on that issue, but we're having this conversation today with Raise Health.

We're designing the initiative very much with these questions in mind, being more proactive before, you know, before these larger scale implementations that you're talking about occur. I think the point you raised about opting in, about complete transparency, about how data's being used, and about giving people the option to not have their data used in any way, that has to be a critically important part. Yeah, if you want to engender trust and get buy-in from the public, I think that's absolutely mandatory

because that trust has been breached in the past. And so it's even harder to earn right now. And I think for good reason. Absolutely. Yeah. What is the regulatory landscape look like? Like what is the FDA doing? How are they looking at it?

How are people like yourself interfacing with them? Like what does that communication look like? What are the barriers to this type of technology that regulators are throwing up? And where are the green lights? I think in the case of the FDA, the FDA is really leaning in to learn and understand how AI could be used

and how they need to be responsibly involved in regulating the application of AI. So there's a branch within the FDA looking at digital diagnostics, for example, or digital health care delivery to know how should that be regulated. The FDA collaborates with or funds institutions like Stanford, for example, with UCSF. We, Stanford, have a Center for Excellence in Regulatory Science and Innovation, goes by the acronym CERCI. So we have faculty at both institutions doing the science that informs regulation. The FDA funds

this. They receive the reports. They're very much involved in the activities of the center. funds this. They receive the reports. They're very much involved in the activities of the center. But the FDA is partnering with or gaining information from those that are doing the primary AI and healthcare work to know how they need to be involved in regulation. Likewise, the Department of Health and Human Services, through the Office of National Coordinator of Health Information Technology, is very interested in how AI is and can be deployed in medical records in the future. So I think the governmental agencies are taking a responsible approach of wanting to understand the landscape and know how they can and should be developing the regulations that absolutely will be needed.

But I'm encouraged. I've gotten to know and learn from some very, very dedicated public servants at HHS, at the FDA, both through their own work and the work that they are engaged with us and other institutions in are gaining the knowledge needed to do responsible regulation. And regulation has to be a part of this. Sure. It's not hard to imagine the many ways in which these technologies are going to kind of upend healthcare. But at the same time, our healthcare system is pretty recalcitrant to change. It's Byzantine. It's complicated. It's confusing. It's expensive.

It's broken in so many ways. And many a person has tried to untie this knot and sort of slinked away unsuccessful in doing this. How can we change this system and make it better? Irrespective of AI or maybe in partnership with AI like we need a better system and incremental changes don't seem to be the way forward because it's so systemic so how are you thinking about that big problem it is a big problem and you know we could spend another series of podcasts and with a lot of other people talking about how we got to where we are right now in the U.S. healthcare delivery system.

We do have, and we talked about this before, a great sick care system in terms of providing ultra-specialized care for severe acute illnesses. acute illnesses. But we don't have a great system for preventing disease and also for allowing a broad-based access to the care that people need. Look, what I think about as a leader of an academic health system is what we need to think about is where we can have beneficial effects, where we can do things that lead to benefit that are actionable. And that is, by definition, incremental. For example, if deploying AI in the interpretation of chest x-rays eliminates or dramatically reduces the 4% of missed significant abnormalities, then we've improved care delivery. And if we show how it can be implemented into workflows in ways that radiologists embrace

it and not push it away, and we talk about that and others do it as well, then we've led by example. So a lot of what we try to do is think about how can we responsibly innovate, study, rigorously study an innovation that we've introduced into the delivery system, and then talk about what works in ways that lead to others adopting it. I think that's a primary responsibility for us as an academic health system. At the broader healthcare delivery system level, if you look at Medicare, for example, moving to Medicare Advantage plans that better align incentives for keeping people healthy rather than, you know, after-the-fact care when people get sick, if there are incentives for

providing in-home care rather than having people go to the emergency department every time they have an issue or a problem. I mean, those are things that I think the Medicare system is doing to make Medicare first better for patients and more efficient. I don't think there's going to be a massive top-down overhaul of the U.S. healthcare delivery system in a way that's implementable. But through a series of these interventions, I think we can and will have a better delivery system. Those are measures that are oriented around acute symptomology, but it's still a long way to go to get into the predict and prevent. But I can imagine with these tools, the diagnostic tools, the wearables, and all of these things, that there can be a more seamless transmission of data to your primary

care physician. And they are kind of in constant communication with all of their patients and are alerted if something is awry. Again, there's privacy concerns with that, but to the extent that these tools can allow us to communicate more seamlessly with our caregivers, I can imagine that would be quite disruptive in a good way. Absolutely, and can allow us to get information about in-home monitoring, for example, and real-time information about when a person has some, with chronic diseases, for example, has some early signs that their health is

about to rapidly decline and to be able to intervene before it gets so severe that they need to come into the hospital. So we do that today with visiting nurses, but increasingly we should be able to do it with electronic monitoring. Again, privacy is a big concern, but those are the types of things. The other thing to keep in mind, Rich, about U.S. health care is that, you know, 70 percent or roughly of the determinants of disease in our country isn't just restricted to the U.S., 70% are socially and environmentally mediated determinants. So the social environmental determinants, social behavioral environmental determinants of health, things like access to the right food supply, behavioral health issues play

a big role. And historically, we haven't done a great job in the U.S. of addressing those social, behavioral, environmental concerns. One of the things that we're trying to do as an academic medical center and school of medicine is, through our Department of Health Policy, is to, again, do what we do well, which is to do rigorous research and then to disseminate the information from that research in ways that it can help drive policy changes and help informed interventions. I mentioned before about replacing cooking utensils in homes with smaller utensils, and that being a positive driver of weight loss.

So things like that, they're incremental, but they do have impact. Right. The idea that your zip code determines your health outcomes in too many ways than it should, that's not going to eradicate the fact that somebody lives in a food desert or just doesn't have access to healthy foods or isn't in an environment that's conducive to moving their body in a healthy way, et cetera. But those smaller interventions, they're still meaningful, but the problem is so much larger than that. And it transcends health care. It's really about our urban landscape, et cetera.

Precisely. And economic disparity. What is the near-term future and the far-term future? I've had futurists, and I had Peter Diamandis in here. He's painting some crazy picture of what it's going to look like. You seem like a much more grounded person. But what does five years look like if these advances continue to move forward? What does 10 years look like? Are you willing to get into the prediction business for five minutes and share a little bit about that?

Or are you going to be too circumspect? for five minutes and share a little bit about that? Or are you going to be too circumspect? No, I can get into the prediction business, but I have to say my crystal ball is pretty cloudy. But, and also I am an optimist, but I've also seen enough reasons for optimism. And I think that's important, particularly in the environment today

where it's been challenging for many people to be optimists. So what do I predict, say, in the next five years? I do think that the field of early detection diagnostics, so being able to have screening tests that provide an early signal that a cancer, signal that a cancer, in the tumors that historically have been diagnosed much later than they should be, pancreatic cancer, ovarian cancer, I think there's good reason to be optimistic that we will have diagnostic tests. There are already some that are available. They will get better and better, and there will be more and more. And therefore, we'll be able to say to a person, based upon your genetic risk profile, you should have this screening panel

looking for cell-free DNA in a tumor from a blood test. You should have this done every year. And those signals will then help to detect those tumors much earlier. That's within our grasp, and I think that that will be rolled out. I also think our ability to do in-home monitoring of conditions will improve, and that we'll have fewer hospitalizations that could have been prevented if we had seen the decline in various parameters that intervened earlier. I think that's going on and will have

increasing impact in the years to come. You know, coming out of COVID, and COVID was certainly a defining event for everyone as individuals and as a society, but it was particularly a defining and horrific event for healthcare providers because we were on the front line throughout COVID. And I gained so much inspiration from the colleagues I worked with every day as we at Stanford and in our region navigated the challenges of COVID. But we also have to be aware that we have a healthcare workforce today that in so many ways is burned out. And I think now that we're, we still have a lot of COVID around, fortunately not many people are requiring hospitalization, but I think rejuvenating the healthcare workforce, it is a resilient group

of people for sure. But we've got work to do on rejuvenating that workforce. And I think that is something that we will accomplish over the course of the next three to five years. We've always attracted people to healthcare and medicine who have extraordinary dedication to the mission, but they don't have an infinite reserve of resilience. And rebuilding that now is particularly important. How do those tools aid with burnout? One thing is, how do we get people away from the computer terminal? How do we reduce the amount of time that a healthcare provider has to spend doing documentation? How do we help a physician with the inbox? It's great.

Patients should be able to ask questions of their physicians directly and in a secure way. But if that's adding on two hours of work after eight hours in the clinic, in addition to documentation, that contributes to burnout. Those are all things that technology is today already beginning to have impact. And I think in the futurist sense, it's going to have even greater impact over the next five years. Getting health care providers, physicians back to what they went into the profession to begin with, and that is to be able to interact directly with people during some of the most challenging times in their lives.

How far away are we from a future in which there's a pod in your house that you slip into every day or once a week? It scans you. slip into every day or once a week. It scans you. It detects everything that's slightly off in your body, sends it to your doctor or prescribes a protocol to address it. Hey, there's a little cluster of cancer cells here. Let's just, let's get that taken care of right away.

That kind of thing that you see in sci-fi movies. Yeah. I don't think that's five years away or maybe not 10 years away. I didn't say five. Is there a future in which that is a thing? Oh, sure. I mean, look,

who would have thought 25 years ago that the smartphones you and I are using are doing what they're doing today? I mean, a few futurists would have predicted that. Few people did and created companies that have been remarkably successful. But most people had enormous skepticism about it. To the example you mentioned, already today, there are centers, they're not covered by and large by insurance, but where you can go in and get a whole body MRI. Yeah, like a DEXA scan.

by insurance, but where you can go in and get a whole body MRI. Yeah, like a DEXA scan. Yeah, exactly. And a DEXA scan for looking at bone, but a whole body MRI to detect whatever. Now, it turns out that those also sometimes detect things that aren't really a problem. Sure, if you go in and do that, they're going to find something. That's right. And then what do you do about it? Yeah.

So there's a ways to go, but as you and I, you said before, our history of putting things in a box and closing the box and saying we're never going to open it, that never works with technology. The general theme here is that there will be some revolutionary changes in diagnostics. There will also be, if we're looking at applications of AI, there will also be revolutions in drug discovery that are going to help to have better therapeutics that really are tailored to specific conditions in individuals. And that's going to improve health and well-being. Where are we in terms of growing organs? I was at a conference, Sanjay Gupta's conference,

and I can't remember the doctor's name. I think she was growing heart tissue. I don't remember. Maybe you probably know who this person is. But it was pretty remarkable what she was sharing and the future that she was painting. And I think that if Uma Valetti can create hamburgers and steaks and chicken fillets out of brewing these cells, that harvesting or sort of fermenting and growing these organs doesn't seem like that far distant from that.

I agree with you, not only growing them, but as one of our faculty members at Stanford is doing, 3D printing them. Yeah, that's what I was thinking of. With living tissue. Yes. It's in the lab today. As to when it gets to patients, it's a bit hard to predict. And we do a lot of things in the lab that then getting them into patients requires many more steps than we thought it would. But it definitely is going to happen. And so, for example,

It definitely is going to happen. And so, for example, in congenital heart disease, that is children that are born with heart abnormalities, we're going to have a lot more options for treating that than we've had in the past. In the past, there are ways that you could sometimes bypass the problem and some very innovative cardiac surgery techniques. Heart transplantation has become more common. But if we can actually build chambers of the heart, 3D print them, and then put them on the heart and have them function, then we've introduced a whole new way to correct congenital abnormalities of the heart.

That conceptually is here today. Just there are a lot of steps between what's going on in the lab and getting it to patients. Well, it feels like a really exciting time if you're a young person to go into, you know, the biomedicine field to, you know, be a med student right now because so many things are changing so rapidly. Like, it feels very different than, you know, even a few years ago. I completely agree, and we have the privilege every day of working with some of those young people that bring so much inspiration and energy and vision to all of us. There's never been a better time to be in the

life sciences than today. Why? Because there has been a convergence of so many different areas of science and technology now being applied to biomedicine. We talked about 3D printing, which began, obviously, in terms of printing objects. Now we're printing cells. That's going to be a whole field. Right, right. The application of AI to the study of protein structure and being able

to predict therapeutics based upon the knowledge of protein structure. There are just so many things that other fields of science and engineering and technology that maybe as recently as five years ago had been considered as completely different from life sciences, now those fields are very much a part of and are being applied in the life sciences. That's what makes this such an exciting time. What are your daily health practices as somebody who knows a lot and is on the kind of forefront of learning about what's new

and what really moves the needle in terms of trying to maintain your health optimally? Like what are the things that you do every day? Or maybe what are some of the changes that you've made in recent years? Well, having lived in California now a little over a decade, it's easy- A little different than Baltimore. A little different than Baltimore, but Baltimore is a fantastic place. But we get fresh fruits and vegetables here, and that's a large part of my diet is consuming those. I'm not a vegetarian. I do eat meat products, but I do it probably fairly sparingly. I go to the gym. I try to get to the gym three mornings a week. I try to get some

cardio exercise every day. I generally, I just started several years ago not eating breakfast. I actually found when I start, I tell you when I started doing it, I started doing it during COVID when everything was so incredibly busy. We would typically start meetings at 7 a.m., sometimes earlier during COVID. So it just wasn't time to eat breakfast. I found myself getting to lunch and, wow, you know, I actually feel pretty good. And so the intermittent fasting has become a part of my routine. I'm not recommending it to others. You just simply ask what I'm doing. And so those are some of the things. Try to eat sensibly and responsibly and get exercise. And you're a cello player. Do you still play?

I am. I do. I do. That was another, that was a silver lining of COVID for me because I didn't get on an airplane for 18 months. And it was, you know, the evenings, there were constant Zooms, but I could practice again. And I also discovered that Zoom works pretty well for music lessons. So I tracked down my teacher from the East Coast and said, are you doing Zoom lessons? He said, of course. And so it became a weekly routine. And then I started playing with some other groups and some very talented young people. And now, of course,

we're back to the real world and it's become more of a challenge to practice, but it has been good. I saw a video of you playing the cello with Condoleezza Rice playing the piano. Condie's an amazing person, amazing pianist. That was wild. One question I always ask every doctor that I have on the show is, if you woke up in a parallel universe and discovered that you were the Surgeon General of the United States, what would be your priority? What would be your mission statement? What would you focus on?

Yeah, and you've had the Surgeon General. I had him, and then I said, now I actually have him. So I can't ask him that question anyway. I think what I would focus on are the things that help us, each of us as individuals, engage with our health and our health care in transformative ways. So making information about our health readily available to everyone, not just a particular socioeconomic group or particular educational level, but making health information available and accessible. I would also focus on the responsible regulation of things that we

know harm health. And already there's being a lot done on that, but a lot more could be done. I would also focus on making sure that we had a healthcare delivery system and a discovery pipeline that focused attention on diseases and disorders that were most prevalent in the society. What would that look like? I'm not sure I totally understand what you're saying. Well, for example, new ways to treat high blood pressure in addition to our existing therapies. New ways to treat, and we're already seeing this, for example, high blood glucose and obesity.

GLP-1 agonists are doing that very effectively. It took a long time to get to those. Some of that was a scientific problem, just working out the science to develop an effective GLP-1 agonist. But others, I think another reason was it just wasn't at the top of the list in terms of where the emphasis was on discovery. So those are some of the things that I'd want to focus on. We all know that if we're having some kind of symptom, the last thing we should do is go online and look at what people are saying on WebMD or just be an armchair physician and try to diagnose ourselves. Of course lots of people do that, there's a temptation to do that, but now we have these LLMs right

at our fingertips. So when we're experiencing something we're gonna want to type it in and see what these LLMs are going to tell us about what we might have and what we might want to do about it. What is your sense of how reliable they are currently in terms of engaging in that kind of behavior? I know you've spent a lot of time playing around with these things. Is that a wise thing to do? Is that something you should avoid doing in the current state? I don't think you should avoid doing it. I think you ought to have a healthy degree of skepticism on what you see. But back to our previous topic of don't put things in a box and shut the lid. First encounter, I mean, I started playing around with large language models when ChatGPT came out in November of 22 or around there. But

the first indication that I had that this was really going to be transformative came from, I was asked to give a talk reviewing my research over the summer at a course, and it gave me an opportunity to update things. And I just went online, and I'm probably best known in my field for the discovery of an inner ear disorder, and I just went to, this was CHAT-GPT, even before GPT-4 or CLAUD or other, you know, large language models. I'm going to ask it about the thing I know the most about. Correct, yeah, about the thing I know the most about. Correct.

Yeah. And just see, and it was really good. I mean, even some of the subtleties it got, when I recognized some of the language and what it was writing, but the description of the syndrome was accurate. There were, at least in the response I got, there were no hallucinations or glaring abnormalities. Now, when I started pushing it with more and more detailed questions, then, yeah, it falls apart.

But I wouldn't shy away from it. I think, back to your original question, I think the so-called Dr. Google has been a good thing. When I moved at Johns Hopkins from being a department chair and a physician-surgeon-scientist to being provost of the university, needless to say, my patient care activities and my research got curtailed. But when I was a very active physician-surgeon-scientist, people would oftentimes come in having found me because of a search they did and a diagnosis they had made, in many cases correctly, of an ear disorder that I discovered and described. So I think people should be using them, but they should have a high degree of skepticism. But the more knowledge someone has about their health, about their well-being, the better it's going to be.

What is the message that you most want the average consumer who's watching or listening this to kind of understand in terms of the intersection between technology, medicine, and the future that is upon us? I think the message I would want people to know is that there's a stronger potential for good than for bad, but we have to keep our eyes open and be very much focused on the bad that could result from the misapplication of the technology. And consumers have to be actively engaged and involved in knowing how information about their health is being used. And healthcare delivery systems have the obligation to furnish that information. And as you were saying before, to give people the option to opt in if their data is going to in any way leave the ecosystem, that has to be done with their full knowledge and understanding of how it's going to be used.

And what is your message to the young, aspiring med student? There's never been a better time. And I think our medical students today very much have that attitude. You know, we continue, not just at Stanford, but across the board, see record numbers of young people apply to medical school, apply to PhD programs in the biosciences to do the types of discovery-based research that are going to drive the innovations we've been talking about today. Record numbers of highly qualified people are interested in these fields. And I think it's because of the opportunities we've been talking about.

Part of me feels like you're in the wrong job. talking about. Part of me feels like you're in the wrong job. You could be this communicator at large around these ideas and be at the intersection between the private sector and the government or the kind of consumer AI watchdogs to guardrail and definitely guide this technology forward. I guess that's what you're doing, but you have a lot of other responsibilities and obligations in your current role. Well, I said before, it's never boring. Look, I derive so much energy from the people that I have the privilege of working with every day, and communication is a part of it. In fact, communication is part of such a critically important part of any leadership role today. I don't think it's ever

been more important than it is today, communication, because there's so many opportunities for, you know, messages to get misconstrued or for things to be not connected in the way that they should be. We have a leadership academy for people in faculty in the School of Medicine who are thinking about, these are generally mid-career faculty who already, you know, proven themselves as accomplished scientists, physicians, other experts, but who think they may want to move into leadership roles. And a lot of that leadership academy is focused on how to be, first, a good listener, and second, a good communicator, and how to really engage meaningfully with others.

I mentioned to you a conversation I had about a decade ago with a venture capitalist who said that he didn't like to focus on anything that purported to change behavior. And another conversation I had with another venture capitalist, I asked for, well, what do you look for when you're evaluating an entrepreneur? Because you're deciding on investing in an idea, but you're also deciding on investing oftentimes in one person or a very small group of people. And what advice do you give to them when you take them under your wing and fund their company?

And what this person said was, don't assume intentions. Focus on situations, realities, but so much time and energy, psychological energy, time, is devoted because we think, well, why did this person say that? Or why did they—maybe they were thinking this. If you have a question, ask the question. If you have a concern, express the concern. But we need more transparency in our communication. I think that was the essential message.

message. And it's something that I've tried to do in being available, but also reaching out when I have a question or I have a concern, expressing it, giving and introducing a dialogue that can lead to a meaningful solution. Well, you're a very effective communicator. And I think being adept at science and medicine doesn't necessarily mean that you're effective at communicating these things that you know so much about. Those are two very different skill sets. And most people in the sciences or most people in general aren't schooled or educated in how to translate their expertise in a way that is meaningful for other people, which is why I think it's so important that people in the sciences who are, you know, kind of at the front lines or the vanguard of new things be better skilled at how to communicate those ideas. I agree with you.

Communication is critically important in everything we do. And the other message that I think we can all learn from is I believe we can all learn to be better communicators. And it's not like you reach a plateau, right? But it's like learning anything, whether it's learning to ride a bike or learning biochemistry. You have to be intentional about it. You have to focus energy on it. You have to focus your intellect on understanding how to communicate and also really looking critically at when you're not communicating well. And I depend upon others to give me feedback on if I haven't explained something.

I depend upon others to give me feedback on if I haven't explained something. And usually it's because we're all busy. It's an error of omission rather than commission. In other words, that we just haven't reached out or there was an opportunity to have engaged people in a different way at a different time that would have led to more effective, transparent communication. The other thing about that is it's helpful for all of us as leaders to be vulnerable, to enter into interactions in a way that we want to learn from others. We're still evolving. I'm still evolving as a leader, as a person, and join me in this journey. That resonates with people. I do believe at the core people are inherently good, and people respond to requests for meaningful engagement and people respond to leaders who come across first and foremost as caring individuals.

Well, if you want to engender trust, that's the primary thing. You have to be willing to be vulnerable. You have to say when you don't know the answer. You have to admit when you got it wrong. And I think those lessons are ones that need to be learned on a health care public policy front as well. You mentioned COVID. There were missteps and mistakes that were made with that that I think denigrated the level of public trust in the sciences and in our health care system and how we kind of advocate what people should and shouldn't do. And we need to rebuild that trust.

That's exactly right. You mentioned before, when we don't know the answer to something, we need to say that we don't know. I mean, remember early on in COVID when we all washed our fruits and vegetables coming home? I will never forget that, yeah. So that didn't prevent the transmission of COVID. We didn't know that at the time. But we do have to be better at communicating uncertainty and at building trust from the public that medical knowledge, any knowledge, evolves as we get more evidence. And particularly when we're dealing with something new.

Yeah, it's a moving target. Things are changing. People are doing the best they can under the, you know, high pressure circumstances of the day. But I think, you know, being able to kind of say, well, we got that one wrong. We should have done this. Now we know better. But doubling down on we were right, you know, is not the way forward. No, I agree with you.

So hopefully that was, I mean, you know, it's like, did we, like for the next pandemic, are we ready for that? Like, it's kind of scary what might happen. It is scary. Did we learn the lessons we needed to learn to be better prepared the next time? Because there will be a next time. There will be a next time.

And I'm not sure. There are people on both sides of that debate. But, and I think the other thing about it, the did we learn is maybe we could rephrase and say, are we learning? Because I think that process is still ongoing. And also, I think some of the working through of the burnout and just the toll that COVID has taken on all of us, that's still ongoing, and we still will have the, we need the benefit of a bit of an interval. Now, the danger is that once COVID really leaves our minds and we won't focus on making the changes we need to make, we have to avoid making, we have to avoid allowing

that to happen. You know, I've heard, for example, that we had a, during COVID, we had an issue with supply chain in, you know, in personal protective equipment. Are we rebuilding the stockpiles in the way that we need to build them? Before COVID, health care delivery systems used just-in-time ordering. You didn't like to keep big warehouses. It was inefficient. Things would go past their expiration date. So you would get things in just in time. Well, guess what?

That works if you can predict your demand. But when you have an unpredictable, you know, perturbation and you run out of supplies, so are we thinking about building up the supplies that we need and doing it in an enduring way? Those are important questions. Yeah. Hopefully, there's smart people who are working on that and thinking about that while we work hard to rebuild trust. Well, look, it was an honor and a pleasure to have you here today. It was an honor for me. I have tremendous respect for the role that you, the responsibilities that you hold and the way that you advocate publicly around the ideas that you care about. And although I said,

you know, maybe you would be better in a different job, I'm very glad that you're in the job that you're in. Thank you, Rich. And yeah, if there's anything I can do to be of service to you, please let me know. It's been an honor. This was a delight. Thank you for the conversation. I appreciate it.

Thanks. Peace. Peace. Plants. That's it for today. Thank you. at richroll.com, where you can find the entire podcast archive, as well as podcast merch, my books, Finding Ultra, Voicing Change in the Plant Power Way, as well as the Plant Power Meal Planner at meals.richroll.com. If you'd like to support the podcast, the easiest and most impactful thing you can do is to subscribe to the show on Apple Podcasts, on Spotify, and on YouTube, and leave a review and or comment.

Supporting the sponsors who support the show is also important and appreciated. And sharing the show or your favorite episode with friends or on social media is, of course, awesome and very helpful. awesome and very helpful. And finally, for podcast updates, special offers on books, the meal planner, and other subjects, please subscribe to our newsletter, which you can find on the footer of any page at richroll.com. Today's show was produced and engineered by Jason Camiolo with additional audio engineering by Cale Curtis. The video edition of the podcast was created by Blake Curtis with assistance by our creative director, Dan Drake. Portraits by Davey Greenberg.

Graphic and social media assets courtesy of Daniel Solis. Thank you, Georgia Whaley, for copywriting and website management. And of course, our theme music was created by Tyler Pyatt, Trapper Pyatt, and Harry Mathis. Appreciate the love.

Love the support. See you back here soon. Peace. Plants. Namaste. Thank you.