The Power of Generative AI in Healthcare with Dr. Robert Pearl, the former CEO of Kaiser Permanente

[00:00:02] Hey everyone, welcome back to the Outcomes Rocket. Thanks so much for tuning back in. It's been quite some time since I've had Dr. Robert Pearl on the podcast. In fact, I think it's been about five or six years

[00:00:15] Today he is with us and he needs no introduction. But hey for those that don't know. Dr. Pearl Let me just give you a quick intro He was the CEO of the Permanente Medical Group Kaiser Permanente. That is from 99 to 2017. In that role

[00:00:31] He led 12,000 physicians, 42,000 staff and was responsible for the nationally recognized Medical care of over 5 million Kaiser Permanente members on the west and east coast He's here today to share the latest update and that is his new book

[00:00:47] It's called Chat GPT MD. How AI empowered patients and doctors could take back control of American medicine Debut as number one on Amazon's new bestseller lists and all profits of this book get this They go to doctors without borders just an incredible person a lot of recognitions named

[00:01:08] One of modern health care's 50 most influential physician leaders. I keep up with his work on LinkedIn his articles Robert so great to have you back on the podcast. Hey, Saul, it's gonna be great to be back with you next time Let's not wait five more years

[00:01:22] I love that. I love that. Let's make it more frequent couldn't agree more Look, I'm so excited about your book But let's just talk about what inspired you to write this book as you know, I published Mistreated why we think we're getting good health core easy wrong

[00:01:41] 2015 I wrote about the system of medicine the problems that happen when care is fragmented Doctors don't collaborate coordinate When you have payments on a Piecemeal a transactional fee for service basis and I had hopes that medicine would be changed

[00:02:02] Lo and behold a couple years later things aren't very different So I asked myself why what's going on? I said there has to be another factor and I researched it and came up with the idea that maybe was the culture of medicine

[00:02:14] What we learn in medical school and keep with us across our entire careers are uncaring Other culture of medicine kills doctors and patients. I thought I found the secret sauce What was going on underneath the surface and lo and behold a few years later not much had changed

[00:02:30] So I get out to ask myself this one I teach the Stanford Medical School the Stanford Business School and that's how I approach Problems when something doesn't make sense to me. I asked what's going on underneath and I said

[00:02:43] What's going on? Maybe the fact that people know what to do. We talk about as value based care But they don't have the tool and then all of a sudden 2020 18 months ago up pops chat GPT generative AI and now suddenly we seem to have

[00:03:02] Exactly the elixir. I've been writing about this for a while Encouraging Apple to build it into their devices AI Intelligence build it into Siri or build it into Alexa How can we as patients and clinicians Have ready access to information expertise then all of a sudden it appears

[00:03:23] And so that's what drove me both to write the book and to ask chat GPT to be my co-author. I Love that. Okay, so you had chat GPT co-author it with you. That's awesome

[00:03:37] That is fantastic glad you did that I began so by putting in 1.2 million words that I had published So it knew my voice knew my thinking Wow And after that I went back and forth probably 200 times in every chapter

[00:03:51] sending drafts the way I would approach you with a medical student or with a resident writing a paper in a book together the Driver of all that was the reality of how fast generative AI is moving. It's doubling in power every year

[00:04:05] It means that by five years from now, it's gonna be 32 times more powerful It's like your car five years from now being as fast as an airplane Unimaginable to the brain, but that's exactly what's happening. And I realized if I took the same two years

[00:04:19] I normally would take in writing a book and editing it and say it's my publisher and all the other Processes that have happened by two years from now anything I wrote would be out of date. They'd be wrong so I thought I had to speed it up and

[00:04:33] That's why I thought believe that chat GPT would be a great co-author But more than that if I'm recommending this technology for medicine I wanted to try it in a different venue and not words in the writing and publishing of a book

[00:04:47] That's fantastic. Not really appreciate that. Dr. Pearl and look there's a lot of people out there And there's different applications, right? It all took the world by storm Jenna through the form of chat GPT in particular There's mixed feelings right around the technology around

[00:05:04] Hey is should I use it? Should I conceal the use of it? I say use it and be open about it because it is the technology that's gonna change the game. What do you think is

[00:05:16] Going what about chat GPT and Jenna? I is going to change medicine for the better I think that Generative AI and again, I like chat GPT because it's fitting well with my book titled chat GPT M&A all the letters flowing together

[00:05:31] But as you say, it's just generative AI could be Gemini from Google or it could be Claude or one of the others dozens of others that are likely to pop up onto the scene the Advantages in medicine are massive

[00:05:47] There are some risks we need to talk about today, but I see generative AI being like the iPhone It's gonna permeate every part of the way that we live our lives It's gonna permeate every part of medicine. It's gonna permeate

[00:06:03] The patient at home. It's why I talk about the AI and Howard patient It's gonna permeate the clinician in the office is gonna permeate the hospital It's gonna be everywhere because it's gonna provide and it does provide not just knowledge but also expertise if you look at the

[00:06:22] Explosion of information across time you can go back to the Gutenberg printing press all of a sudden rather than books being simply that of the Royal elite you had it available to the middle class all sudden you have universities you have libraries

[00:06:39] You trans you go forward to the 20th century the Internet Now suddenly on your computer You can access information click on links find articles in ways that you never could before Then you have the iPhone you carry this in your pocket in the 21st century

[00:06:55] The generator is gonna be the next leap. But what's different about it is That it conveys expertise So what I mean by that is you can paint a picture not using chat GPT per se but Dolly But these are all derivatives of generative AI

[00:07:10] It looks like a Rembrandt. You can program a computer even though you never had an AI course You can write a song that sounds like Drake even though you don't play an instrument you'll Expertise and the expertise will be the doctor and a lot of the technologies

[00:07:28] Happened over time has been to improve medical care provided by clinicians But it's also gonna be the patient and I want to stress two things First of all, the real power is the triad It's the empowered patient with the dedicated clinician and generative AI and

[00:07:46] That combination is gonna be far more powerful far more reliable far more accurate than any one piece alone and Second I want to stress to listeners of viewers That I am NOT talking about the current version GPT for or any of the other current versions that exist

[00:08:06] But at the rate that it's increasing very soon we're talking a three to five year time frame That's what I'm speaking about. But having said that don't wait This is the time to experiment to try it to figure out what's possible last week

[00:08:22] I was on a different podcast and I was talking to the host After the show and she said to me my husband took a fall a couple of months ago skiing His arm was over his head. He slid down the mountain about a hundred feet and

[00:08:33] She doesn't work well and it's painful what's going on? She knew I was a skier and a doctor so she figured maybe I could help advise her what was going on And I do know I had a pretty good sense of what was going on

[00:08:45] But I said don't let's wait once you put any information into chat GPT She wasn't even using the first version No less GPT for put all the information in place and see what you get

[00:08:57] And a couple days later she's emails back and she says my gosh, this is amazing It made the right diagnosis a rotator cuff tear It recommended the right treatment the MRI and it highlighted that surgery often is needed

[00:09:11] I should consult an orthopedic surgeon because delay could be problematic and lo and behold I did That's exactly what the doctor said had I waited a couple more months He might not have been able to fully reattach the muscle on the tendon that had detached from the bone

[00:09:26] This is what we have today again, no one should use it without Consulting a clinician but they'll provide expertise even in its current form Love that. No, this is fantastic And I love the idea of the triad the patient the physician

[00:09:44] Jen AI and even I would expand it to say the caregiver right like beyond that whether it be a nurse or an RT or Whatever. We've had a lot of conversations Dr. Pearl on the podcast with leading companies in the JI space from physician notes to

[00:10:02] Payments, what are your thoughts around some of the technologies that are? Like really existing today and what's coming in the future? First let me say that I think we should stop talking about in quotes AI We need to separate it into the different kinds of AI

[00:10:19] There's rule-based AI where clinicians put in place algorithms the way we learn in medical school and residency Computer programmers write it. It's very good at giving you a consistent answer

[00:10:30] You know exactly where the information came from but it can never be better than the people who created it. It's really only minimal AI Then you have around the year 2000 introduced neural networks deep learning now You're looking at what I label and people call narrow AI

[00:10:48] this is we take 10,000 mammograms 5,000 which show cancer 5,000 of which are normal or benign disease and the algorithm and the tool itself finds 30 40 50 70 100 different slight differences between these two data sets and is able then to assign a probability factor to each and

[00:11:09] Be able to make a diagnosis about 10% better than clinicians. There are 600 of those today But they're called narrow AI because they only take on a small problem you do mammograms. That's great You can't do chest x-rays or brain scans or anything else with it generative AI is

[00:11:24] completely different generative AI is Taking all of the information on the internet all the textbooks all of the journal articles and putting it into place So that it can answer any question and that is what the power is going to be

[00:11:45] Of this it's gonna allow that patient at home to Better manage their chronic disease think about it We have wearable devices now that are very accurate Very small little devices that can measure blood pressure pulse blood oxygen blood glucose go down a list of physiological

[00:12:02] Parameters that it's able to do and what do we do with that data? absolutely nothing and why did we do absolutely nothing with it because the patient can't interpret it and

[00:12:14] Because the clinician doesn't want at all clogging up their electronic health record or having to spend a massive amount of time And what's the patient gonna do? They have a hundred readings of blood pressure 80 of them are normal 13 or abnormal and seven of them are uncertain

[00:12:29] Do they need more care the next visit is three months from now with the doctor should they call the office? What do they do? This is where expertise is needed and the technology based upon what the clinician puts in for the expectations could

[00:12:43] Answer that question. So going back to what you're asking about the AI that currently exists most of it using generative AI Accomplishes I'll call more in quotes administrative tasks, but there are quite a number of tools right now that are

[00:12:59] Quickly becoming AI scribes able to listen in on conversations with patients and create the electronic health record That's needed because right now clinicians spend way too much time staring at their screens not looking at their patients There's certainly a lot of tools that have been designed

[00:13:16] Most of those are the narrow AI variety not the generative AI to battle insurance companies and be able to code differently and higher Many of these are gonna be very successful companies They're not as interesting to me as this notion of how do we transform American health care?

[00:13:30] How do we move truly to a value-based health care system? How do we address the fact that there's 400,000 people die every year from misdiagnosis another 400,000 who remain and permanently disabled

[00:13:44] From medical from an error in this diagnosis. What about the quarter of a million people who die from a Preventable medical error in a hospital. What about the one in four individuals who have a misdiagnosis while they're in patient?

[00:13:59] What about the hundreds thousands of people who have poorly managed chronic disease? Diabetes hypertension asthma we manage hypertension 60% of the time at best it could be 90 95 percent. We manage diabetes less than a third of the time This is what is interesting to me and worries me

[00:14:18] Because it's not that clinicians aren't smart. It's not that they're not dedicated This is not something negative about them They just don't have the time to do these things and that I believe is why we see

[00:14:32] 60% of physicians being burned out an equivalent or maybe even higher number of nurses How do we help them and augment it? This is not a tool that Replaces doctors and nurses

[00:14:44] This is one that fills in the gaps that augments the care that creates the time that we need If we apply it appropriately and that's what has me so excited about this introduction of generative AI and why I wrote chat GPT MD

[00:15:01] Thank you, thank you so much for that. Dr. Pearl when people pick up your book. What should they expect to gain after reading it? There are five parts to the book Okay, the first three parts Explains what is generative AI?

[00:15:18] How did this tool come about? How is it different than the other AI's I briefly mentioned it here and more importantly What are the applications? How will it be used in hospitals? How will be used in doctors offices?

[00:15:30] Very importantly how will be used by patients at home and what's likely to happen not just in the first five years But then the first ten years we said it'd be 32 times more powerful after five years It's gonna be a thousand times more powerful after ten years

[00:15:44] We're gonna take all the information that we use in clinical practice coming off the electronic health record Coming of all the monitors that we have in the operating room Coming off of how human beings use robots in an operating room or pass catheters in cardiology

[00:16:01] Laboratories and it's gonna learn how to predict what we would do next and duplicate it It's gonna have a massive amount of opportunity And so that's the first three parts and I want to get people excited by them part four is about all the problems

[00:16:14] This is not a technology like any technology that doesn't have I'll call negative outcomes or our privacy security bias misinformation the threat of job loss and Then part five in many ways to me. The most important part is going to be leadership

[00:16:33] Because this technology no matter how powerful isn't gonna happen unless we can help people overcome their fears and For clinicians as you mentioned earlier using this technology Creates risk creates first a risk on one hand that a Mistake is going to be made

[00:16:53] That something will be done to harm a patient. There's nothing that as clinicians We fear more than inflicting harm on a patient, but there's another whole side that needs to be addressed Which is really going to be how does this impact us both professionally and personally?

[00:17:13] And for a personal standpoint if you think about this technology, what are we really trying to accomplish? The American healthcare today is unaffordable With over half of people saying they can't pay their out-of-pocket cost today

[00:17:28] What this technology can do is to keep people healthier if we do the things I mentioned earlier on chronic disease and There's a lot of data that said would be able to reduce the number of heart attacks strokes cancer renal failure peripheral amputations by 30%

[00:17:44] What was it due to American medicine cost it would drive it down? Dramatically and make health care far more affordable But what would it do potentially to clinicians? It would erode their income they're paid on a piecemeal fee-for-service basis

[00:17:59] That's what we said, and if they have less work to do they might make less money That's pretty risky and threatening. You got a family you got a house The key is moving from the fragmented from the piecemeal fee-for-service

[00:18:13] Basis to some form of prepayment pay for value capitation call it what you want Because now suddenly it's just as valuable to prevent the heart attack as to reverse one It's just as effective to avoid a colon cancer and having to do a major resection followed by chemotherapy

[00:18:31] This is the opportunity that exists and so leadership will be necessary because as problematic as health care is today the human nature is to persist and continue that which you know over that which is unknown the Comfortable even if it's uncomfortable The common even if it's uncomfortable

[00:18:53] Over the future uncertain and that's where leadership needs to make change happen I often say what leaders do is they make things happen that otherwise would not when people have

[00:19:03] Crossed that threshold they never would go back the other direction and that's what I believe will happen with generative AI if we can apply Effectively in medicine for comfortable empowering patients, and if we can work together, that's why I want to start now

[00:19:19] Figuring out how are we going to train patients around this how we can help them to use it reliably and effectively Because we don't want already doctors are too busy clinicians are too busy

[00:19:30] And if we don't use it effectively then people just aren't gonna do it and the technology will be Sounding great, but it will not have the impact that I believe that it needs to happen to improve quality to make care convenient To reduce burnout into lower costs

[00:19:47] Yeah, thank you so much for that Robert Thanks for breaking it down for us from setting the foundations on AI to the different elements around it the risks and ultimately leadership Folks if this books not on your shelf already then order it now

[00:20:02] We're gonna put a link in the show notes so you could access it on Amazon. So make sure you check that out I imagine Robert it's available to ship as well as a digital Copy of it. I'm sure all of them are available

[00:20:14] There's a digital copy if people want it in that form that go to your Kindle There's a you can have it shipped if you're a prime member. They'll ship it pretty quickly

[00:20:24] Hopefully overnight and later this month. I'll be recording an audio version for anyone who would like to go to that as well But right now yes, it's available on Amazon as you said, it's number one

[00:20:35] That's new books list Stanford just published its new books list. It was number one there as Well and anyone who wants more information go to my website Robert Perlmd.com There's a lot more information on all these topics and I encourage viewers and listeners

[00:20:51] That if you buy the book the profits will go to a good charity doctors without borders. Let me know what you think Let me know whether you're excited about this or whether you're fearful of this if you're a clinician Let me know how you're gonna use it

[00:21:03] Some recent data that says 40% of physicians plan to use it next year in their clinical practice I think we're at the dawn of a new era I call it health care 4.0 and I welcome feedback the broader the

[00:21:19] Conversation the more diverse the opinions the faster will be able to transform American medicine and I believe make it once again the best in the world Amazing dr. Pearl. Thank you so much for that. I too am very excited about the

[00:21:33] Stage in history that we are right now. And so for everybody listening take dr. Pearl up on that He also records a podcast if you haven't had a chance Fixing health care and medicine the truth. He's got a newsletter called monthly musings on American healthcare

[00:21:50] Contributes regularly to Forbes. We're gonna leave all those links in the show notes, but the most important one is the book So make sure you take action on that book

[00:21:58] Dr. Pearl, thanks for being with us today. Thank you so much. And I thank you on behalf of chat and chat GP TMD