Subscribe to The Podcast by KevinMD. Watch on YouTube. Catch up on old episodes!

We sit down with physician executive Robert Pearl to explore the revolutionary concepts presented in his latest book, ChatGPT, MD: How AI-Empowered Patients & Doctors Can Take Back Control of American Medicine. Join us as Robert delves into the transformative potential of generative AI in reshaping the health care landscape. Discover how this innovative technology can empower patients and providers alike to reclaim control from corporate interests, ensuring a future of higher quality, accessibility, and affordability in American medicine.

Robert Pearl is a physician executive.

He discusses his book, ChatGPT, MD: How AI-Empowered Patients & Doctors Can Take Back Control of American Medicine.

Our presenting sponsor is Nuance, a Microsoft company.

Do you spend more time on administrative tasks like clinical documentation than you do with patients? You’re not alone. Clinicians report spending up to two hours on administrative tasks for each hour of care provided. Nuance, a Microsoft company, is committed to helping clinicians restore the balance with Dragon Ambient eXperience – or DAX for short. DAX is an AI-powered, voice-enabled solution that helps physicians cut documentation time in half. DAX Copilot combines proven conversational and ambient AI with the most advanced generative AI in a mobile application that integrates directly with your existing workflows. DAX Copilot can be easily enabled within the workflow of the Dragon Medical application to bring the power of ambient technology to more clinicians faster while leveraging the proven and powerful capabilities used by over 550,000 physicians.

Explore DAX Copilot today. Visit https://nuance.com/daxinaction to see a 12-minute DAX Copilot demo. Discover clinical documentation that writes itself and reclaim your work-life balance.

VISIT SPONSOR → https://nuance.com/daxinaction

SUBSCRIBE TO THE PODCAST → https://kevinmd.com/podcast

RECOMMENDED BY KEVINMD → https://kevinmd.com/recommended

GET CME FOR THIS EPISODE → https://earnc.me/ntVbg5

Powered by CMEfy.

Transcript

Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast, get CME for this episode by clicking on the CME link in the show notes. Today we welcome back Robert Pearl. He’s a physician executive, and we’re going to talk about his new book ChatGPT, MD: How AI-Empowered Patients and Doctors Can Take Back Control of American Medicine. Robert, thank you so much for coming back on the show.

Robert Pearl: Thanks Kevin, it’s great to be back with you.

Kevin Pho: So Robert’s been on multiple times on the podcast. Go to KevinMD.com/podcast to search for his name and prior episodes to hear his story. But today let’s jump right into your new book, ChatGPT, MD: How AI-Empowered Patients and Doctors Can Take Back Control of American Medicine. So how did this book come together?

Robert Pearl: The book started last fall. But let me remind the viewers and listeners that all profits from the book go to Doctors Without Borders, which is a wonderful charity, and hopefully will generate a lot of interest in the United States and around the world.

So it began last fall, when I at that time was using ChatGPT 3.5, the first version, and it switched to GPT-4, and I could see the massive advance that happened in the technology. And it was very clear to me that there were applications in medicine, not just in the traditional way that AI is used, which is to support doctors, but actually to empower patients to be able to take greater control of their health.

And I was looking, and everyone was writing about the shortcomings, and they were right, the problems of the ChatGPT original and the GPT-4 were actually there. It did hallucinate, and it often didn’t have all the most recent information available. But this was exponential growth. It was doubling in power every year, and I could see that by five years from now it would be 30 times more powerful than today, and at that rate a thousand times by 10 years.

And we were missing the ball on what this type of generative AI technology would do to change and improve American medicine for both doctors and patients. And I decided I wanted to educate people about that, to write my third book. But I realized that if I went through the traditional way in which it was done, that’s a two-year process between writing and editing and printing and reviewing, and I decided I wanted to shortcut the process, and I’d make ChatGPT my co-author. And that was what I undertook last fall.

Kevin Pho: All right, so you obviously have a unique perspective as a physician, you’re a physician executive, you’re a former CEO, and you have a blend of perspectives. So when you were using ChatGPT and you noticed that huge jump in computational power from 3.5 to four, of all the things that struck you as it relates to health care, just talk about, say, the top few that struck you in terms of how it can affect us as physicians and how it can affect patients.

Robert Pearl: First, for listeners and viewers, think about how much improvement happened so fast. So think about the fact that this tool actually passed Step 2, the licensing exam that we all take as doctors, but it barely passed it in the first version, and six months later it scores in the top 10 percent of people taking the exam.

Or look at the fact that people looked at messages written by clinicians versus ChatGPT. This is work coming out of the University of Arizona, and they found that not only was ChatGPT better able to answer patients’ questions through email or through text, but it was four times more empathetic. And then they gave this tool as a test to the physicians in Boston, and they found that it gave better answers to complex problems.

So this is the evolution. So how do we look at this in terms of the opportunity once there’s empowerment of patients? Well, first of all, as you know, Kevin, very well, six in 10 Americans have chronic disease. This is the biggest problem. The diabetes, the hypertension, the asthma. It leads to heart attacks, strokes, kidney failure. We know what’s going on, and we do a poor job across the United States of controlling it. Talk to clinicians about it, they blame patients often for not doing the things that they’ve been told, but maybe we don’t have the right tools.

And that’s what ChatGPT first of all can do. Rather than treating a chronic disease every three to four months in a doctor’s office, imagine, particularly if it’s linked in with wearable devices through what’s called plugins, which by the way were released last fall, these are the advances that are happening now, patients can see how they’re doing, how effectively they’re managing hypertension, we control only 55 to 60 percent of the time across the United States, or diabetes, that we control less than half of the time for patients with type 2 diabetes. Now they can be managed on a daily basis with reminders, information about prevention, improvements in lifestyle. So that’s the first set of opportunities that exist.

The second set, and you’ve led the way in conversations about this, Kevin, is burnout in doctors. And I read a fascinating study that looked at burnout across the world. And when I talk to physicians, they look at the corporatization of American medicine as the underlying problem, the prior authorizations that they have to go through, the tremendous number of administrative tasks, the electronic health record that, as you well know, is predominantly designed around financial means of documenting information for insurance companies for payment.

And I assumed that we’d see that American medicine, physicians in the United States, were twice as burned out, particularly in primary care, as the rest of the world. But no, we’re right in the middle, better than four countries, slightly worse than four others, right in the middle of burnout around the globe.

I said, how is that possible? And I looked at it and thought about it, and I said, the problem could well be chronic disease. I mean, if you as a physician see someone with an acute problem, you take care of it, you prescribe a medication, and hopefully a few weeks or a few months later it’s all better. But with a chronic disease, as soon as you take that patient on, for the rest of that person’s life you’re going to see them three, four times a year.

And I realized a lot of burnout is demand supply imbalance. The demand on doctors is just too much, and it’s been estimated that it would require something like 27 hours a day to do all the right things. And now we have a tool that can empower patients, if clinicians work with them. So now patients can know if their problem is significant or not, they can find out what’s the most likely diagnosis that they have and the treatment. They can take care of a lot of problems.

And I believe that it could take 20 percent of what’s done in doctors’ offices today and do it in the home, assuming that clinicians work with patients. And then of course there’s a lot of opportunity in our hospitals to improve outcomes. We know that, as an example, nurses see patients every four hours, lots of things can change in that interim between rounds. ChatGPT can monitor that.

And most recently OpenAI released what’s called memory. These were the two big problems that happened six months ago, again, this is the rate of improvement. Hard to enter data, that’s where plugins came in last fall, and the machine wouldn’t remember what happened in the past, so everything was a first time encounter. Now these two things are starting to be available.

And we could reduce the misdiagnoses in the United States that account for 400,000 deaths, to 250,000 preventable medical errors that have been well identified in the past, and hundreds of thousands of deaths from heart attacks, strokes, kidney failure, systemic infection. These are the opportunities that are there. It’s massive. It’s what medicine has wanted for so long. This is the evidence-based medicine, higher quality, easier, more convenient access, and greater affordability.

Kevin Pho: So we are speaking at the end of March, and this episode will go out probably mid April. So as it relates to directing patients to these large language models to help manage your chronic disease, we have GPT-4, we have Google’s Gemini, we have other large language models. Are you confident enough in their accuracy to recommend patients to use these devices to answer some of their medical questions, to help manage some of their chronic diseases?

Robert Pearl: Not today. I would not recommend that patients rely on this in place of clinicians. I do believe that it can assist patients to understand their diseases. I think it can assist them to gain medical information. I think rather than them trying to figure it out alone, being able to access one of these tools in order to be able to figure out whether their problem is life-threatening and needs to be addressed in the middle of the night rather than just going back to sleep, so there’s some opportunities that are there.

But no, the tools are not there. But as we just said, within five years it’s going to be 30 times more powerful, and I’m absolutely convinced that by five years from now, yes, we will be able to rely on them. But that doesn’t mean we should wait.

Because one of the things you learn, and in writing this book with ChatGPT as my co-author, I started by putting in 1.2 million words that I had published, and then I probably had two, three thousand interactions with ChatGPT. And that’s the key, it’s how do you enter the prompts. The more information you provide, the better your questions, the more rounds you’re going to go through, the better answers you’re going to get, the more accurate they’re going to be. And that is something I think that patients can start doing right now.

But what I want to encourage is clinicians to start using it. Since I announced the fact that my book ChatGPT, MD was going to be coming out, it will, by the way, be published on April the 9th, be available on Amazon at that time, I’ve heard from lots of doctors who were writing to me to tell me, because I had cautioned people against using it now, how accurate it was in their experience, that they had pretended to be patients, entering data, information, getting advice, and how closely the information they got today reflected their own.

So I want to be clear on two things. Number one, no, it’s not ready yet today. It needs to be updated in terms of the information, needs to have improvements in the so-called hallucinations, making up information. There are still problems to be worked through, but they will be. But no one should think of today’s problems as tomorrow’s future. This technology will revolutionize, transform American medicine, and I’m hoping, Kevin, that clinicians will lead the way, because if they don’t lead the way, someone else will, and that never is good for American medicine.

Kevin Pho: So in your ideal world, what are some specific ways physicians can lead the way in terms of integrating these AI tools into daily clinical practice? Give us a hypothetical scenario of that ideal interaction, or a specific way physicians can lead the way.

Robert Pearl: Well, today, the first thing they should do is get very familiar with the technology. They should try it to figure out how it works, they should enter information in it, they should see how it occurs, because they can’t teach patients unless they are skilled at applying it.

I think they can take patients with chronic disease and explain to them how they can apply the information inside the tools of today around lifestyle medicine. We know we could diminish diabetes, we know we could avoid complications from chronic disease with better lifestyle medicine actions. And we’re talking about diet, exercise, relaxation, all the different things that have been shown to diminish the complications from chronic disease. Clinicians can lead that way.

They can work with patients who are doing home monitoring, to be able to put the information into their computer systems and get information out of it, not to actually act on it per se, but to be able to notify the clinician when the problems are becoming worse as opposed to better. Why does it take so long to be able to get a chronic disease under control? It’s because we manage it so episodically. That kind of information could be monitored currently utilizing one of the generative AI tools.

They can use it to provide post visit information to patients, and do so not just in English but in different languages. So there’s a lot of, I’ll call them small things, that could be done today, but that’s all in preparation for what’s going to be available tomorrow.

I think that we, the national group, should be looking at ways of figuring out how to educate patients once this tool becomes exponentially stronger, to learn how to do the prompts, to figure out how to connect the wearable devices, whether that’s going to be a blood pressure monitor, a pulse monitor, a blood oxygenation monitor, all the other information, to be able to put that into the generative AI system so that it can allow a daily management of patients.

How are we going to put in place all the information? Think about genomics. How much data is there? And I don’t know clinicians who can remember all the pieces. There’s just too many parts to the human DNA genome. But for a generative AI tool that’s not a big problem, that’s not a lot of information on a comparative scale.

So these are the ways to be looking at the future. Today is how to learn to be comfortable with the tool, but tomorrow is the way to be able to embed it into our practice of medicine, both in terms of being able to have the patient utilize it and the clinician to rely on it, to look for additional diagnoses if a physician’s stumped, wants a broader sense of a differential diagnosis, wants to have the most up-to-date protocols from national medical organizations. These are opportunities to obtain it.

And I want to stress to listeners and viewers, it’s much better than what’s available today on Google. You go to Google, what do you have? A whole series of links. And if you’re a clinician, you can probably figure out which ones to rely on, which ones to disregard, you may be able to figure out how to get to the right answer that you’re looking for. Patients can’t figure that out. It’s why they come to the doctor’s office with hundreds of pages of printouts coming out of Google, because they can’t figure it out and they see a lot of information, much of which is misleading. Generative AI is better than that and can help sort it out.

And I mentioned earlier plugins. What will be able to be available at some point in the future are plugins created by physicians that would then go into the generative AI systems of patients, which will tell them the parameters for which clinicians are concerned, the direction that they expect improvements in chronic disease to go, and when there’s deviation from that that is significant, be able to notify the patient, notify the clinician, and allow more rapid changes in medication without having to schedule a visit and without taking a huge amount of the doctor’s time.

Kevin Pho: We’re talking to Robert Pearl. He’s the author of the new book ChatGPT, MD: How AI-Empowered Patients and Doctors Can Take Back Control of American Medicine. Robert, let’s end with the main message from your book that you want readers to come away with, as well as some take-home messages for my audience.

Robert Pearl: The first message is for patients who are listening in, and that is that this technology is coming, and you should become used to using it, figuring out how to enter information. The more you enter about your medical conditions, the more specific it is, the better your answers are going to be. Don’t use it today, use it for an information source, not for a clinical source, but be prepared for that in the future.

For clinicians, the message is much more complex. The first piece is that you too need to become skilled at utilizing this technology. The second is, you have to become comfortable empowering patients. That’s not been the tradition of medicine. It’s always been, in the doctor’s office, in the hospital, clinician knows best. And we’re going to have to start to rely on and trust our patients. They already are managing their finances, their travels, their trips. Medicine is not that much more complex, yes, it’s more dangerous, the greater risks that are sitting out there.

But the biggest piece is going to be recognizing that to maximize the use of this technology, to reverse the stagnation in life expectancy in the United States that hasn’t gone up in 20 years, and the unaffordability of medical care, we’re going to have to make two changes.

And one is that we’re going to have to work more collaboratively with each other, putting together groups of individuals to be able to provide the kind of support to patients that they’re going to require once they start utilizing this tool. And we’re going to have to move from fee for service to a form of prepayment, I’ll use the word capitation. And that is because the power of a generative AI tool is not by doing more volume, it’s by keeping people healthier, it’s by avoiding chronic disease, avoiding the complications from them.

And as clinicians we need to have value, and by that I mean income derived from keeping people healthier, from avoiding excess visits to ERs, avoiding hospitalizations, making sure the right care is done the first time without misdiagnosis, without medical error. And to do that we are going to have to be willing to have the courage, to have the leadership, able to transition from the current reimbursement system today to the future.

I know it’s risky, I know people are worried about what’s going to happen if something should go wrong, if patients don’t listen, if colleagues don’t do their part. But I think that this is the future of medicine. We’ve been talking about it now for decades. ChatGPT is the catalyst that will allow it to happen, because finally we have a tool that will allow patients to be able to take greater control of their health, and as a consequence of that to stay healthier, avoid chronic disease, avoid its complications, and get the best medical care as quickly as possible at a cost that they could afford.

Kevin Pho: Robert, thank you again for coming on the show and sharing your perspective.