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

We sit down with guest Andrew Wilner, a neurologist, to explore the transformative impact of artificial intelligence on medicine. Join us as we discuss how AI is changing the role of health care professionals, the tasks it can take over, its limitations, and the enduring importance of the patient-physician relationship.

Andrew Wilner is a neurologist.

He discusses the KevinMD article, “Will artificial intelligence replace physicians?”

Our presenting sponsor is Nuance, a Microsoft company.

Together, Microsoft and Nuance are leveraging their rich digital technology and advanced AI capabilities to tackle some of health care’s biggest challenges. AI-driven technology promises to revolutionize patient and provider experiences with clinical documentation that writes itself.

The Nuance Dragon Ambient eXperience, or DAX for short, is a voice-enabled solution that automatically captures patient encounters securely and accurately at the point of care. 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.

Physicians who use DAX have reported a 50 percent decrease in documentation time and a 70 percent reduction in feelings of burnout, and 85 percent of patients say their physician is more personable and conversational.

Discover AI-powered clinical documentation that writes itself. Visit https://nuance.com/daxinaction to see a 12-minute DAX Copilot demo.

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/1InS9Z

Powered by CMEfy.

Transcript

Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast, and get CME for this episode by clicking on the CME link in the show notes. Today we welcome back Andrew Wilner. He’s a neurologist, and today’s KevinMD article is titled “Will artificial intelligence replace physicians?” Andrew, welcome back to the show.

Andrew Wilner: Thanks, Kevin, it’s a pleasure to be here.

Kevin Pho: So Andrew was on, I think, about two and a half years ago. For those who don’t remember that far back, just briefly share your story and journey.

Andrew Wilner: Well, I am currently an associate professor of neurology at the University of Tennessee Health Science Center in Memphis, Tennessee, so that’s my day job. For many, many years my other job has been that of a writer. I write medical news articles for Medscape and other publications, and I do a lot of interviews, actually like this, for RadioMD, ReachMD and Medscape, because I like to know what’s new and what’s going on. I also have my own podcast called The Art of Medicine, where I interview people I like to interview: physician authors who’ve written books, either nonfiction or fiction, or people who are doing interesting things, usually outside of their regular jobs.

And so, because I do like to stay on top of things, this issue of artificial intelligence came up. There were a lot of articles, and suddenly there are hundreds of experts. I don’t know how there are so many experts when this thing’s brand new. I even interviewed an expert, an engineer, on my show about artificial intelligence. But I thought I should weigh in on how it’s going to affect the medical profession.

Kevin Pho: So you’ve done multiple podcast interviews, and you’ve written for multiple publications. What are some common questions you get asked about artificial intelligence and its intersection with physicians?

Andrew Wilner: Well, I think the most poignant question came through my website, AndrewWilner.com, where I usually entertain questions about locum tenens, because that’s my other interest. I did that for many years and wrote a book about it. A woman from Australia wrote to me, actually I think she left me a voicemail, saying that her son was considering medical school, but did I think it would be a waste of time now that there would be artificial intelligence, and should she encourage him to do something else? And I thought, wow, that’s really something. I think in the back of everyone’s mind there’s that question. When I wrote my article, the issue was: Is this going to put me out of business, and can artificial intelligence do what I do? And I think the answer is actually very complex.

We all use artificial intelligence every day, right? On your phone you can dictate a message, and somehow your voice turns into words that are typed. I mean, that’s pretty amazing. It can, however, put people out of work. Every doctor’s office used to have a back room full of transcriptionists. I remember I used to dictate into a tape recorder, and I would leave the tape on the counter, and the next day there would be four busy ladies in the back room typing it and complaining about people’s accents or mumbling and having to go back and review. I would not encourage anyone to be a transcriptionist. I think that has gone by the wayside.

Similarly, my dad was a CPA, a certified public accountant, and I remember as a kid saying, “Dad, what are you going to do today?” He had a client who had a big men’s clothing store, selling suits and sport coats and things like that, and he said, “Well, we have to shut down the store because we’re going to take inventory.” This sounds incredible to somebody who was born in the last 20 years, right? They shut down the store and manually counted every piece of clothing and labeled it, so they would know what they had sold and what they hadn’t sold. And they hired people to do that. Well, those jobs are gone, because we do that with barcodes now, right? Every store manager knows at the end of the day exactly what sold and what didn’t sell, because it’s all automated with the barcode. So there’s no question that computers are pretty powerful and can change our lives and can put some people out of work.

But my take on it was that computers are really good for information retrieval and information storage. And in medical school, right, we spend a lot of time learning information: anatomy, the names of drugs, their half-lives, their indications, their side effects. That information now is much more easily retrievable than it used to be. I have an app on my phone, PDR.net. I used to have a giant book, right? You’re probably old enough to remember the giant PDR books that were on every nurse’s station and had all the information. Now it’s on an app. So AI is good for that. AI is good for recognition. AI is being used in pathology to look at a slide: Is this a benign mole, or is this a malignant melanoma? AI is pretty good at that. AI can read EKGs. That’s been going on for 30 years, right? The EKG comes out of the machine and it’s got an interpretation. However, that interpretation is not always right, and it has to be overread by somebody who actually knows what they’re doing.

Why? Well, one of the issues is artifact. I go back to learning Fortran when I was in college, when computers were as big as my office here, and there was a saying, GIGO: garbage in, garbage out. One of the problems is that we can say we can monitor patients’ blood pressure or oxygen 24/7, but it’s not always right. And if the raw data isn’t right, well, the output isn’t going to be right, and computers aren’t that good at determining whether the raw data is useful. So that’s a very, very big limitation. This is going on right now in other areas, right? If I tell my accountant how much I earned and that information is wrong, my accountant’s calculation of how much I owe in taxes with their computer program is going to be wrong. So getting the data right in the first place is very, very important, and I think that’s a big limitation that’s been skipped over in AI.

The other obvious limitation of AI is that AI is not a person, and doctors do a lot more for patients than simply follow algorithms. In other words, if you come in with an infection, an algorithm will tell me the best antibiotic and the dose. The computer will tell me that. But then I have to convince you that that’s the best thing to do: to take it, how to take it, how long to take it and how it’s going to interact with your other medications. And you have to have confidence in me that what I’m telling you makes sense, so you don’t try herbal therapy or just do what your neighbor did.

So I think artificial intelligence is a great tool, just like a calculator. When I was in school we had to memorize multiplication tables and division and all of that. Now you just pick up the calculator, and every now and then I want to do a quick calculation in my head and it’s like, you know, I’d better use the calculator because I’m rusty, right? I can guess pretty well, but the calculator is always right. It’s a machine. But I don’t really want a calculator when I go to the doctor.

So what I wrote in the essay is that physicians who rely only on algorithms will, I think, be replaced, because computers are going to do algorithms: take data and come up with a solution, or apply the guidelines. They can remember all the guidelines a lot better than I can. I can look them up, but the computer will have it in a split second. But dealing with the patient, the laying on of hands, is historical. Why have there been physicians for the last 5,000 years? They didn’t do much. They didn’t have antibiotics, they didn’t know anything, right? But there was always a physician. Why? Well, because with the laying on of hands, physicians knew, or had a pretty good idea, about prognosis. Where is this going? Am I going to die, doc, or am I going to be OK? I need to know. And how can you alleviate pain, how can you reassure me, how is this going to affect my family and my job? I think it’s going to be a long time before AI is able to manage the human considerations of illness and disease.

Kevin Pho: Do you see a scenario where AI has any role in enhancing that human aspect, whether it’s taking over some of the electronic medical record work that doctors have to do in the exam room, so they can better focus on the patient? Are there any scenarios where you can see AI in the future perhaps enhancing that relationship with patients?

Andrew Wilner: Well, I think it would be great if AI could flag things that sometimes get buried in the history, like this patient’s father died at age 40 of colon cancer. That’s important for the doctor to know, but a lot of the specialists zooming in and out don’t take family histories anymore, so there ought to be a little red flag there. Similarly, screening for depression can be done very simply with just a few questions, but it gets buried in the EMR. A little flag should come up: This patient had a positive score, something to think about, something to be aware of. Prior hospitalizations for the same thing. I think AI could really do a great job if it had access, and of course this has been discussed, if it could somehow not only read the fields but also read the text of the doctor’s note. The AI that we have now tends to just look at fields.

One other area: Very recently there was an article that suggested AI could interpret electroencephalograms, which I did a one-year fellowship to learn how to do, as well as people can. And it’s like, whoa, am I going to be out of a job? Well, no, because it still makes mistakes. But one of the things it could do: We sometimes monitor patients 24/7 with brain wave studies, and that is very time-intensive to review. Frankly, it’s humanly impossible to take a pile of these, or a computer disk full of four patients who were monitored last night, and review it in an obsessive-compulsive way if you’re a human being. But a computer with the right algorithms and the ability to separate artifact from brain waves, if it could do that, would be like me having a full-time, highly trained assistant without actually hiring one. So I think there’s a potential for AI to really contribute.

But again, I always think of GIGO. That is huge. The AI has to be given a sample that is meaningful and accurate, or what comes out the other end may look great, but of course there have been a lot of stories about ChatGPT where people say, “Oh, tell me about this,” and the computer just makes stuff up, like saying Napoleon was active in World War II, stuff that anybody would know is wrong, anybody except the computer. So I think AI has a long way to go, but it’s certainly going to be part of our future tools.

Kevin Pho: I think that with AI, or ChatGPT specifically, it’s been out for about a year now and it’s already made so much progress. So I think it’s a reasonable assumption that a lot of the errors we’re talking about now, in terms of reading EEGs and EKGs, will significantly decrease with the next generations of AI. Now, what can physicians do to maintain their humanism, to maintain the doctor-patient relationship that AI can’t replicate going forward? Because AI, if it gets more accurate, is going to take some things off our plate, but it’s never going to replace the relationship. As physicians, how can we maintain that relationship with this improving AI around us?

Andrew Wilner: Yeah, I think that’s a great question. Well, one approach is to improve the system, right? Having more time, being less rushed, all of those things. If you and I are going to have a conversation and you tell me, “Hey, Andrew, I’ve got five minutes,” it’s like, OK, that kind of puts me off, right? If you said, “Hey, Andrew, I’ve got all day, you just tell me what’s important to you,” all of a sudden the whole tone of our conversation is different. Well, obviously doctors don’t have all day for each patient. The good ones actually pretend they do, and I’ve heard patients say, “He talked to me like he had all day.” Well, of course he didn’t have all day, but he didn’t look at his watch, he didn’t get interrupted, he didn’t fiddle with a computer, he didn’t leave the room. There are system things that can be done to enhance the doctor-patient relationship, and I would say that in general the opposite is occurring, right? Doctors are more rushed than ever, they’re interrupted more than ever, and they’re asked to do things that are way below their skill level, like typing. So there’s a lot of low-hanging fruit in the system to fix.

I would say the other thing is a mental awareness, a situational awareness that what I really have to offer this patient is not only my knowledge but my heart and my soul and my caring. One of the things I started to do intentionally: When I get to the patient’s room, my instinct is just to barge in there and get going, because I have a lot of other things to do. So I try to remember to just stop at the door and breathe for a second, let everything stop, take a breath, and then go in like it’s a new day and a new person, and leave everything behind. That literally only takes about three seconds, so I think that’s a recommendation other people can try without messing up their day.

It’s that situational awareness, and being hands-on. Touching patients used to be a lot easier. Now there’s always concern that the touching is somehow inappropriate, so I think that has created a bigger distance, because the laying on of hands is important. When someone actually touches you, there’s a connection. If it’s unwanted, that’s a problem, but hopefully when you’re with your physician, you want that human connection. I know when I go to my doctor, I always tell him irrelevant things about my life that have nothing to do with the case. I don’t know, he may find it annoying, but I want him to remember me as a person, and not as the guy who came in to get his creatinine checked, or whatever it is: “Oh yeah, he was rambling on about this and that.” Because he’s a human being too. So I try to make that connection. I think the human connection is, as you suggest, extraordinarily valuable.

Kevin Pho: What do you anticipate the second year of AI will bring to our lives as physicians?

Andrew Wilner: Well, one area where it might help is writing letters. I do a lot of writing; that’s what I do. I asked ChatGPT to write me a letter the other day. Like a lot of people in the world, I want a raise. So I told ChatGPT, “I want a raise,” and I gave it a few specifics, and it wrote this beautiful four-paragraph letter that frankly would have taken me all day, and it had some things in it that I wouldn’t have thought to put in, like what a great job I’ve been doing, not just inflation. Anyway, it had to be edited, but it was an interesting exercise. So I think it’s a way to get started.

I don’t think it can replace writing, because usually when I write, I don’t really know what I’m going to write until I sit down and write it. So I don’t want to skip that step, the step of actually making yourself write. That is when you figure out what you want to say. So if you just let ChatGPT do it, you kind of miss out on that whole maturing thought process. On the other hand, it is a great way to get started and not forget things.

I do use a writing and editing program called Grammarly. I don’t get anything from Grammarly for recommending them, but I’ve used it for several years and I’ve found it very helpful. Sometimes I’ll use the same word over and over again without realizing it, or I forget a comma, and it picks that up. And often it makes suggestions that are wrong, and I think that’s the key thing: None of this can be taken at face value. It’ll say, “Oh, you should change this word to that,” and it’s like, no, no, that’s not what I’m saying. So you have to be smarter than the AI, for sure, but it is a useful tool.

I think it would be great to plot vital signs, to plot creatinines. Our EMRs, and we could have a whole session on that, and I’ve written about EMRs, are an embarrassment, right? They could be so much more useful, and maybe AI could help get them up to speed, so that they’re useful tools instead of giant paperweights.

Kevin Pho: We’re talking to Andrew Wilner. He’s a neurologist, and today’s KevinMD article is titled “Will artificial intelligence replace physicians?” Andrew, let’s end with your take-home messages for the KevinMD audience.

Andrew Wilner: Well, I think the bottom line of my essay and our discussion today is that physicians really separate themselves from AI by being thoughtful, kind and compassionate. And I think physicians who are growing up today, who are going to be using AI more and more, need to really focus on their people skills and make themselves really valuable to their patients.

Kevin Pho: Andrew, thank you so much for sharing your time and insight, and thanks for coming back on the show.

Andrew Wilner: Oh, it was my pleasure, Kevin.