Come with us as we talk to the clinicians making AI part of their daily workflows, and using it to build deeper connections with patients.
Welcome back to Trust But Verify, the Evidently Podcast. I'm Andre, and I'm joined by my cohost, Doctor. Kai Romero. If you don't know Kai, Kai is an emergency and hospice doctor, and we both work for Evidently. If you don't know Evidently, Evidently is a clinical data intelligence platform.
Andres Krogh-Walker:We read everything in the patient chart. That means the scan documents, the faxes, the imaging, data trapped in third party HIEs. We index it, we turn it into insight, and we deliver it back into the EHR in the workflows that clinicians rely on every day, usually when they need it most right at the point of care. We recently hosted the first Scaling Health AI Summit, where we had the great pleasure and honor to bring together 18 leaders from, I think, some of the most forward thinking health systems in the country for two days of roundtables, panels, and closed door conversations about some of the biggest challenges facing them in their roles and facing their teams today. During the summit, we were able to grab some time with a handful of these leaders for one on one conversations.
Andres Krogh-Walker:And one of those was our guest today, Doctor. Dieter Sumerauer. And the thing I loved about this conversation was the enthusiasm that Dieter has and the deep thinking that Dieter does about his practice of medicine and his approach to informatics in his role as now an associate CHIO at Rady Children's in San Diego. And Dieter's seen the whole arc of technology as it's evolved in healthcare from being an early adopter of the electronic health record over twenty five years ago in his early private practice through to his role now overseeing the rollout and adoption of technology and AI tooling at Rady's Children's. And what was fascinating to me was that at every step of that evolution, he was finding ways to make the technology work for him.
Andres Krogh-Walker:And using the technology to help build deeper and deeper trust and relationships with those patients, rather than just replacing cognitive thinking. I think I say that every episode is my favorite episode, and this definitely was too. But without any further ado, here's Doctor. Dieter Sumerauer.
Kai Romero, MD:So I want to start off by asking you about your origin story. Well maybe not your entire origin story, but your journey specifically from training in pediatrics to full time clinical care to associate, CHIO. Was it a career path that you intended? Did you see a kind of north star, in informatics, or did you kind of follow curiosity that developed along the way?
Dieter Sumerauer, MD:Yeah. No. It was more sort of, a path that developed out of necessity. And then, you know, you you go through your career and at some point you sort of look back upon what you've done. And I was like, oh, I've been doing informatics for the last fifteen years or so.
Dieter Sumerauer, MD:Back in the early two thousands, I started my own practice. And, I'm proud to say that I was one of the first practices in Northeast Ohio, private practices to have been fully on an electronic health record. I taught myself how to wire rooms, how to create wireless networks that were secure, and, you know, I was on a software back then called Soapware. Some of some of the listeners might remember it. Not a very sexy program.
Dieter Sumerauer, MD:It wasn't, what we were used to today, but it got the job done. It looked like an open chart when you looked at it. On the left, you had the patient's history and on the right, you had the note of the day. Yeah, I did some very sort of, I wouldn't even call it programming, but creation within that program on how to store an immunization record. I'm a pediatrician and it just sort of evolved.
Dieter Sumerauer, MD:Then when I went back to work, so I sold my practice to the organization where I had trained, sort of started my leadership journey with them. As we developed our primary care network or our primary care institute, I was asked to be a regional medical director for what turned into informatics. Yeah, it was just a very exciting time helping our practices develop the tools that they needed to deliver care in a more efficient manner. I rolled out one. I helped roll out one EMR.
Dieter Sumerauer, MD:Allscripts was the first EMR that we rolled out to our practices. Before I left, I helped the organization transition from Allscripts to Epic. And then I was tapped to be the associate CHIO at Children's Hospital of Orange County, CHOC. And that's where I am now.
Kai Romero, MD:So early 2000s, not a time when many physicians were like, yes, let's do electronic health record. You know, it felt like I knew many people who were either really struggling with it. I knew physicians who were saying, You know what, I'm going to retire rather than have to learn that skill. What was it about the EHR that compelled you to be an adopter in that way on really the cutting edge?
Dieter Sumerauer, MD:Yeah. Well, I think I've always been an early adopter when it came to technology. Yeah. I had an Apple IIc when I was in seventh grade and I always was looking at technology as how can this make life a little bit easier. And I saw the first iterations of electronic health records as making our lives as physicians easier.
Dieter Sumerauer, MD:Communicating with one another, keeping track of the record in much more facile way, right? Rather than having this wall of charts in my office, I had a bank of servers. And at the time, it wasn't even like sure. I mean, it evolved. Right?
Dieter Sumerauer, MD:At first, it was a couple desktops that networked together and I mean, this was really early on. Yeah. And it was just I mean, it was a small practice. I mean, I started it from the ground up, so it wasn't lots of patients on day one. But it was a lot of fun.
Dieter Sumerauer, MD:And I saw I really saw the power of what the electronic record electronic health record could do for us. Mhmm. When I started the practice, one of the local community newspapers came out and interviewed me. And I have the article at home still. Have it framed.
Dieter Sumerauer, MD:And one of the things they asked me, they said, you know, what do you find to be one of the benefits of keeping your electronic your records electronically? Because it was it was novel at the time. Right? It wasn't really something that many were doing. And really what I had found at this point, I'd been using it for about oh, I guess it had been about a year and a half at this point.
Dieter Sumerauer, MD:And I said, you know what I've really developed is the ability to review with my patients in real time the information they shared with me. So literally what I was doing at the time is I was typing the note in front of the patient in the room, and many times before I finished, I would sort of turn the screen and I would say, now let me just review this with you. Let me make sure that what I've typed in is accurate. And it helped me not just make sure that the record was accurate, but make sure that my diagnostic path was accurate, that I really had heard what the patient was there to tell me about Yeah. And I wasn't missing anything.
Dieter Sumerauer, MD:And I think that was really one of the very first things that was kind of an Because, you know, I grew up in a handwritten medical record world. And when I first went out into practice, we were dictating. And you would get the note back the next day and then you would review it from memory to make sure that it was accurate. Well, electronic health records sort of got rid of all of that because it was all done in real time.
Kai Romero, MD:Yeah. It says something really interesting about you though, Dieter, that I think one of the challenges that we currently have is how much we many clinicians perceive of the medical record as like this place for internal conversation between doctors and with legislation that opened those doors thinking about how they might need to change their documentation. But it sounds like you have always thought about it through the perspective of how is the patient perceiving of this record that I'm keeping about their care, which is pretty unusual. I don't know that everybody in a longitudinal way thought that way about record keeping.
Dieter Sumerauer, MD:You know, we're taught in medical school that the medical record is a medical legal document.
Kai Romero, MD:Yep.
Dieter Sumerauer, MD:And I think you're right. I think sort of what's drilled home is the fact that this is something that other people are going to look at, that other people are going to review. But who's it about? Yeah. Right?
Dieter Sumerauer, MD:This is a story about the patient. And who better than to help edit that story but the person who it's about? Yeah. I I've never thought of it that way. I don't I I I don't know.
Dieter Sumerauer, MD:Maybe the colleagues that I trained with thought similarly. I didn't really think of it as a novel idea at the time. I've always thought the medical record was not just a medical legal document, but it was a communication device. It was communication device between me and the other caregivers who were taking care of of the patient, to make sure that we all understood the story that was there. As I think about it now, I think about sort of how much more voluminous that has become as the electronic health record has become the derogore of of charting.
Kai Romero, MD:Yeah. I mean, do you remember like the little panic that everyone had once like all the medical records were gonna be available to patients? Like I can't imagine that you had shared that panic because it sounds like you've always been sharing your medical records with patients.
Dieter Sumerauer, MD:Yeah. That's interesting. I I hadn't thought about it in that way. There was that panic. Right?
Dieter Sumerauer, MD:I worked to help champion the open notes initiative at my previous organization, and I'm working on that in my current organization. I think the data today is is pretty pretty strong. Know, the the more a patient knows about their medical history, their medical journey, the more they understand what they've talked about with their physician or their APP, the better care they receive. The more involved they are, the healthier they are. Yeah, I guess I've never thought that that was rather pioneering.
Dieter Sumerauer, MD:But yeah, there was a panic around that and I've never felt that panic because it was their story to share with me.
Kai Romero, MD:Yeah. I think it's more unusual than you realize. So between treating kids in your day to day work in urgent care primarily now and managing big IT initiatives across a large and growing network. What brings you joy in your day to day?
Dieter Sumerauer, MD:You know, I went from primary care and having those long standing longitudinal relationships with my patients and families. There's nothing that that can can top that. You know, that's that's an ROI that you you can't put a dollar sign on. But I will tell you something that comes pretty close and and maybe giving it a run, that feeling a run for its money is having colleagues now who come to me and say thank you for bringing this technology forward. Thank you for introducing me to this.
Dieter Sumerauer, MD:As we've rolled out some of our AI technologies at my organization, I've seen some real change. I've seen some real change in the way physicians practice. I've gotten comments. This has put the joy of practicing medicine back into my life.
Andres Krogh-Walker:Mhmm.
Dieter Sumerauer, MD:There's one provider who actually added to her schedule. Not more patients a day, but she was seeing patients two days a week, she's now seeing patients three days a week because the technology allowed her to finish the work she had, the administrative duties she had in seeing a patient and be able to go home and still have that quality time with her family
Kai Romero, MD:Yeah.
Dieter Sumerauer, MD:And add an entire day. That's pretty phenomenal. And that I think all of us in informatics I like to say that I've always wanted to give my colleagues the ability to practice the highest quality medicine in the most efficient manner. I think most of us in informatics feel that same way. I'm actually seeing that now and that brings real joy.
Kai Romero, MD:I've heard you mention this kind of a couple of different ways and in a couple of different scenarios where you, while being very tech forward and very excited about new technology, I've heard you many times loop it back to the concept of trust. And specifically you've mentioned that having a really good understanding of the patient chart, especially in the urgent care setting where you know it's your first time seeing the person, that that establishes trust with the patient very quickly. What is the role that technology has both in establishing trust with patients and establishing trust with providers?
Dieter Sumerauer, MD:I mentioned those longitudinal relationships with my patients and families in my primary care practice. Now as I see patients in the urgent care, I thrive on It it gives me energy. But now I have, you know, fifteen minutes. And some of the summative AI technologies that we've seen come to market really give you the power of going through a patient's chart very quickly and summarizing that patient's medical history much like a resident would on rounds. And being able to walk into a room and even just show the data to the again, you know, turn the screen to the patient and family.
Dieter Sumerauer, MD:This is your story. Is this right? Do I have an understanding? That establishes such trust and such rapport with the patient and the family that it's like the speed of light in terms of those relationships that I built over twenty five years. I can now build something close to that in a relatively short amount of time by knowing the patient's past medical history.
Dieter Sumerauer, MD:Now there's still that that connection that you wanna build Yeah. With the family. You wanna spend that time doing that, but these technologies and these technologies support that as well. Because by giving me the time I otherwise would have spent reviewing the patient's chart to get a full sense of who this person is and serving it up to me in seconds, I can take that time and spend it with the patient and the family to build a relationship in a very short amount of time just by interacting with them. I think for physicians, giving us the ability to understand what's in that patient's chart.
Dieter Sumerauer, MD:You know, it's interesting earlier today I was giving some thought to this and I thought when I as a physician review a patient's chart, when any of us as physicians review a patient's chart, we review it with bias. I was at a conference earlier today and we were talking about this. And this is actually I didn't think of it at the time, but this is one of the takeaways.
Kai Romero, MD:Yeah.
Dieter Sumerauer, MD:The AI sort of gets rid of that bias. Right? Because it it it takes the patient's chart and it turns it into data. It's just objective. There's no bias.
Dieter Sumerauer, MD:There's no like, oh, I I have to look for this and so I'm maybe not gonna look for that. And it's not it's not conscious. Right? No. We just we all come at the chart from our own perspectives of of what we know.
Andres Krogh-Walker:A lifetime of experience.
Dieter Sumerauer, MD:Exactly. Or or a certain subspecialty might sort of bias me to look for something over something else. Right? The AI doesn't do that. And so I think that as as my colleagues start to use these summative AI technologies more, they're gonna start to see that more and more and that will build that trust as well.
Dieter Sumerauer, MD:Right?
Kai Romero, MD:Yeah.
Dieter Sumerauer, MD:To say, oh, I didn't realize that this was going on. Now there's not that's not to say that there there can't be biases biases built in by the AI as well and we need to always look for that. But it just it put a very different perspective on on this for me today after this meeting. Yeah. And I really have spent the afternoon thinking about it.
Dieter Sumerauer, MD:The other thing I wanna add because we're talking about the the generative suburbative technologies as I like to call them, there are also the ambient generative technologies. Right? And they have changed the way that I practice. Yeah. I told you I would type the notes up when I was in the room with my patients.
Dieter Sumerauer, MD:I haven't touched a keyboard in two years since starting to use these technologies. And now that time that I spent typing, and although I was I like to think I was very good about making eye contact and typing without looking at the keyboard, there was still this third thing going on in the room. Right? It was me typing and interacting with this computer. I do use the mouse, you know, to reconcile meds and allergies and look at vital signs, but I'm not typing in the room and I can actually just focus on the patient and that interaction.
Dieter Sumerauer, MD:I walk in. I ask permission to record the conversation, use this new technology. And for the most part, ninety nine percent of patients are very receptive to it. The kids love it. Yeah.
Dieter Sumerauer, MD:And and going back to what I said earlier about reviewing that history and I said I would type up the patient's note in the chart and then review it with them. Now I can hit the button and it just populates on the screen and and the the kids are just floored by this. Right? Even some of the parents are like, woah. Look at that.
Dieter Sumerauer, MD:Right? And you read it back to them or they read it with you and they're like, wow. The AI just summarized everything I said. And then if there's something wrong, we correct it. But it establishes again that trust and rapport that I'm actually listening and I'm doing it in real time with you.
Dieter Sumerauer, MD:That's huge. Yeah. That's actually huge. Never would have thought twenty five, twenty six years ago that this is what I would be talking about today.
Kai Romero, MD:Yeah.
Dieter Sumerauer, MD:Right?
Kai Romero, MD:Yeah, absolutely. I wonder too about how many really great doctors who were unwilling to consider how technology might augment their work left the profession or didn't keep working as long as they might have. It kind of felt like in a lot of ways, the best predictor of whether or not you'd still be practicing medicine well beyond the potential expiration date was whether or not you were technologically inclined or at least open. Maybe you didn't have to be the earliest adopter, but you had to be open.
Dieter Sumerauer, MD:Well, and you used my favorite word. I don't even know if you realize, but augmented is my favorite word in this, right? And I actually like to say that we're using augmented intelligence because what these artificial intelligence tools are doing is they're augmenting my intelligence and what I'm able to do with and for my patients. And if we look at it from that lens, I think it takes on a completely new meaning for us. I'm not a technician.
Dieter Sumerauer, MD:I'm not an automaton. And the AI, the artificial intelligence is never going to be able to take the place of the relationship that I have with my patients. But it can augment the intelligence that I have and and what's the percentage of our brains that we actually use, right? Yeah. I mean, it's it's going to augment the intelligence I have to provide that next level of care.
Kai Romero, MD:Yeah.
Dieter Sumerauer, MD:Right? To really provide the best possible care in the most efficient manner doesn't mean that that efficiency is going to I'm going to see more patients a day or I'm going to leave the office earlier. It means I can actually spend quality time with my patients in the room and not have to do some of these other administrative duties that have really bogged us down. We all know that with the advent of the EHR was the advent of more tasks and more boxes to check and more things to do. But this is where these technologies can really step in and augment what we're doing.
Kai Romero, MD:Well, and I've heard clinicians say too that by really forcing the charting to happen in real time, it's actually opening the door for the clinical thinking to happen in real time.
Dieter Sumerauer, MD:For someone who actually typed the note during the visit and all of a sudden now I'm relying on the AI to record the conversation, Interesting. I would leave the room to go and look at it and edit it at first. Right? And many times I was left thinking, oh, what did they say? And there were a couple times where, you know, it's it's artificial.
Dieter Sumerauer, MD:Right? So it doesn't get everything right. Right. And there were a couple times where I'd have to go back in the room and knock, I I thought you said you gave ibuprofen. Did you give ibuprofen or Tylenol?
Dieter Sumerauer, MD:And the AI had written down Tylenol and was ibuprofen. I mean, little examples like that. Yeah. So I think actually, like, generating that note in real time in front of the family and getting back to what I was starting to do twenty five years ago, which is review the note in real time, actually gives you that chance to do the critical thinking in real time and involve the patient Right. In that critical thinking and in some of the decision making, which I think is so important Totally.
Dieter Sumerauer, MD:In health care. Right? I mean Yeah. We wanna be involved. It's it's my body.
Dieter Sumerauer, MD:It's my life. I wanna be involved and and part be part of at least driving the ship where we're going. As a patient, I don't have the knowledge that my physician does about what's going on. But as a doctor, I want to hear what my patient has to say. I want to know, you know, we talk about social determinants health and and and all of these things that we've put labels onto but what does all of that mean?
Dieter Sumerauer, MD:Right. It's really just about knowing your patient. Really knowing your patient and being able to understand the predicament, the human predicament that they're in to provide them the best quality care.
Andres Krogh-Walker:And I think that's the trust gap with a lot of patients too is for most patients, think their own medical record is a black box. And having an expert there to translate that for them and show them what's going into it is for some people I think probably the first time they ever had that happen when they're sitting with you and reviewing the actual surface area that their medical records getting committed to. It's a mystery.
Dieter Sumerauer, MD:Yeah. Right? That's terrifying for people. Yeah. I I often think about how someone who has no how a lay person who has no understanding of the health care system can actually navigate the health care system.
Dieter Sumerauer, MD:I think it's hard enough for those of us who are medical professionals. I've had family members who needed healthcare myself. Trying to get answers or trying to understand where to go next or what the process is is so hard. So at least it's the start. This isn't the panacea of course, right?
Dieter Sumerauer, MD:But it may be the start of getting us to think differently about how we leverage the medical record to provide care and support our patients in their health journeys.
Kai Romero, MD:Yeah, I wanted to talk about something I also heard you begin to talk about earlier today and I really wanted to hear the rest of your thinking about it. Which was this concept of what standard we hold AI technology to and what standard we hold clinicians to and how that's different and what that looks like.
Dieter Sumerauer, MD:I think that many of us have this position that the AI needs to be perfect. And the AI can't make any mistakes. Yeah. The reality is that's not true for any of us. We talk a lot about high reliability medicine and in healthcare we don't want to make errors.
Dieter Sumerauer, MD:We really and we shouldn't. Right? And we shouldn't be excusing them. I am wholeheartedly on board. But why are we asking our artificial intelligence to be better and more accurate than we are?
Dieter Sumerauer, MD:I'm not saying that we just open the gates, turn it loose, you know, let it do what it does. That's not what I'm saying. What I am saying is we need to, again, use it as an augmentation. Yeah. We need to review it.
Dieter Sumerauer, MD:We need to critically think about what it's telling us. We need to to look at it, look at it again. Sometimes it's going to be right. Statistically, there are going to be times when it's going to be wrong. And that's where we have to be in the loop and we have to be regulating it.
Dieter Sumerauer, MD:So I don't want to be misconstrued. I'm not just saying, Oh, we don't have to like you know Yeah. Yes, we have to push it. Just like we have to push our human beings, right? Earlier today, someone made the comment that we will sometimes hire someone who is less capable just to get the work done.
Dieter Sumerauer, MD:Well, okay. If we're going to accept that from a human, why aren't we trying to use AI to get some of that that those tasks done Yeah. That none of us really like to do
Kai Romero, MD:Yeah.
Dieter Sumerauer, MD:To help augment the work that we're doing.
Kai Romero, MD:It's really interesting because I think what I've come to think about that standard that we hold ourselves to, and I have two thoughts about this. One is that we actually do in medicine hold ourselves to an impossible standard. Errors are not allowed ever. And if they happen, they're a source of great shame and sadness and self flagellation. And that's something that inculcated into the culture of medicine.
Kai Romero, MD:And I think there's been a lot of work to try and improve that, you know, to try and make it so that M and M conferences aren't just opportunities to point fingers and assign blame. But that's an inherent part of medicine. So I sometimes feel like, well, doctors are really just applying the same terrible metric that they've inherited to their technology. But the other thing that you said that brought up something as well for me, which was, you know, and this is going make me sound like an AI apologist, but a lot of the issues that we see when we see AI doing things are not unique to AI. So we'll talk about hallucinations and charts as if humans don't.
Kai Romero, MD:We'll talk about precision errors as if humans don't commit those. These are errors of cognition. They happen if you're a robot. They happen if you're a person. They actually happen a lot less often if you're a robot.
Kai Romero, MD:But I think one of the things that we were talking about today is like is the standard perfection or is the standard human output in a medical record? Because if the standard is human output in a medical record, AI can already have fewer hallucinations, fewer precision errors and more accuracy. If the standard is like perfect recall with never any errors, then like yeah, there's going to be issues. But I think it's a problem of the culture of medicine that seeps into how we then evaluate technology to my mind Where you know you fax something to the wrong location. That's a HIPAA violation, that's a medical error and the hammer of Thor comes down.
Kai Romero, MD:There isn't a way to like even make an administrative task or a clerical task that should feel low stakes in a medical environment and not have it potentially have serious, serious ramifications.
Dieter Sumerauer, MD:So I've had a flood of thoughts.
Kai Romero, MD:I want to hear all
Dieter Sumerauer, MD:of them. And I don't know if I can. I don't know if I'll get to them all. The first and foremost one is, of course, we think that way because there is a life on the line. Absolutely.
Dieter Sumerauer, MD:There is a human being, a person who is at the other end of what we're doing. So of course we should hold ourselves to that standard.
Kai Romero, MD:Yeah.
Dieter Sumerauer, MD:There there's there's no excuse not to. It reminded me of a of a talk I heard some time ago and the speaker will forgive me for not remembering who said these words. But we have a process in health care where we punish ourselves for mistakes. Understood. But we don't give ourselves time to examine when we've gotten it right.
Dieter Sumerauer, MD:We just move on. It's just expected. When I heard that, it gave me pause to think. Because part of me says, well, absolutely. I should be getting it right.
Dieter Sumerauer, MD:So what am I gonna like? Right? But when you're training and you're not given the opportunity to actually take a pause and look at what you did to make that situation right, how can you repeat it? Right. How can you learn from it?
Dieter Sumerauer, MD:What I really would like to ask, and maybe this is the same thing, is that we not make errors. And this is where AI can
Kai Romero, MD:help. Mhmm.
Dieter Sumerauer, MD:Right? Because to your point, it's not gonna replace us. It's just going to augment us. And I'm I'm gonna be a broken record about this because I I really, feel very strongly that this is about augmented intelligence. And this is how we're going to get to a place of zero error.
Kai Romero, MD:Yeah. And I think ultimately, you know, the sanctity of the responsibility that we've given and the promise to first do no harm has to hold the most weight in terms of what we're doing day to day. But the challenge is that we are imperfect. We do make errors. And many of them, I would argue, are not errors because we haven't thought enough, we haven't thought hard enough, we haven't cared enough.
Kai Romero, MD:But many of them are because in fact, the thing we've been asked to do is impossible. And the responsibility we take on as clinicians is how could I not have read every note in that chart and figured out about this thing that happened twenty years ago? This was actually a real story from one of our users, IBC filter. Just never, nobody ever wrote about again, happened twenty five years ago. If I had known that story, if I had known that piece of information, I would have cared for them differently.
Kai Romero, MD:Or if I had known this risk factor, I had known this thing. And I think the challenge that I see is in holding the clinician responsible for that information that it is truly almost impossible to know without the assistance of a tool that actually helps you bring those things forth. That's where it feels like it's just a punishment vortex as opposed to like a useful exercise in figuring out how we can do this differently. I think that's what I've seen more often than okay structurally what have we done to give the clinician enough tools to do their job well? That is more often, I think, more fruitful conversation around how we avoid medical errors than it can be a really individualistic conversation.
Kai Romero, MD:That I think is, I mean look, sometimes people are negligent. Sometimes people really are not providing the standard of care. But it's far less common than not having the resources or tools to adequately diagnose and treat based on a variety of factors that are outside of the clinician's control.
Dieter Sumerauer, MD:I think what I heard you say is that we're asking imperfect beings to be perfect. Right. I'm about process. When I teach my medical students, I tell them do the same thing you do in the same way every time you do it. I'm old enough to know what a Rolodex is.
Kai Romero, MD:Yeah. Me too.
Dieter Sumerauer, MD:And I tell them if you do that, you'll have this Rolodex in your brain And when you get to that part of the exam, how to do a neuro exam, you'll be able to pull that card out and you'll be able to read it. You'll be able to do that neuro exam because in your training, you did a neuro exam on every single patient, every single time you saw them, just as an example. Think about what we've done in our ORs. We've created improved processes. We have timeouts.
Dieter Sumerauer, MD:We write left and right on limbs. We mark which limb we're going to operate on, which side of the body we're gonna operate on. All of these are added tools or again augmentations Mhmm. To our thinking that improve process. And what is AI but process?
Dieter Sumerauer, MD:It's a review of a medical record all done in process. Yeah. So I think there's a lot of fear about AI where we need to have more understanding what's behind it. I think we also need good governance and good regulation and good discussions about where we're headed, and how we're gonna use it. So there's there's certainly lots to talk and think about.
Kai Romero, MD:Yeah.
Andres Krogh-Walker:I think there's a a really interesting and sometimes scary question about accountability too Because in the same way that AI can have less or a different bias than a human, the consequences for bad doctor versus bad robot are very different.
Dieter Sumerauer, MD:I think it's I think it's part of the tension. I'm gonna add a little caveat to that as well. The doctor who's using the tool is not the individual who created the LLM. Yeah. And isn't behind any potential bias that could be in the model.
Dieter Sumerauer, MD:Yep. So there's someone else who probably isn't even a physician.
Andres Krogh-Walker:But is in the room in a way.
Dieter Sumerauer, MD:Who is in the room in a way through the model they've created with any potential biases that they've brought into the model unwittingly. Yeah. Right? So I
Andres Krogh-Walker:think that
Dieter Sumerauer, MD:it's multifactorial, that fear we have. I think it's also a fear of not knowing. Not knowing where the model came from. Yeah. For many, not knowing what's it doing?
Dieter Sumerauer, MD:How's it doing it?
Kai Romero, MD:This is something I know that you and I have talked about before, but it seems like there's a constant balancing act that informatics leaders have to do between physician autonomy and then system wide standardization. So how do you balance that?
Dieter Sumerauer, MD:You know, I am often heard saying for as many different physicians there are, there are that many different ways to practice medicine. It's hard. For me personally, I have to come back to who are we doing this for?
Andres Krogh-Walker:Who
Dieter Sumerauer, MD:is our customer? Many in my field would say the customer is the end user. I actually see the end user, the clinician, as the conduit to the real customer, and that's the patient. There's still room for the personalization, for those interactions we have with our patients and families. There's room for that in a world of standardization, of standardized processes that get us to better outcomes.
Andres Krogh-Walker:You're listening to Trust But Verify,
Dieter Sumerauer, MD:the
Andres Krogh-Walker:Evidently Podcast. You can find us online at evidently.com/podcast, and you can find us anywhere you subscribe to podcasts. See you next time.