Advancing Clinical Workflows With Responsible Healthcare AI with Dr. Graham Walker, Co-director of Advanced Development for Permanente Medical Group and ED physician at Kaiser Permanente San Francisco Medical Center

[00:00:02] Welcome to the Memorial Health Care Delivery Podcast. Through conversations with industry leaders and innovators, we uncover ways to simplify how patients and care teams navigate complex care delivery. Hi everyone, this is Dr. Jamie Colbert, Chief Medical Officer for Memorial Health and host of the Care Delivery Podcast.

[00:00:25] I'm joined today by Dr. Graham Walker, who is Co-Director of Advanced Development for the Permanente Medical Group. He is a ED doctor at Kaiser Permanente San Francisco Medical Center, and he's also the creator and co-founder of MDCalc.

[00:00:42] So really excited for this conversation today because in addition to all of these amazing accomplishments that Graham has, I also have a personal connection with Graham as we both attended medical school together at Stanford.

[00:00:56] Graham why don't you just start off by having you introduce yourself and tell our viewers how did you get interested in healthcare technology? Yeah, thanks Jamie. I'm glad you mentioned the Stanford connection.

[00:01:07] I was gonna say that's the most important reason we went to med school together a long time ago. It's a great question. So I'm 43, I'm almost 44. And so I only say that to explain that kind of the internet started really when I was

[00:01:21] in high school, very much like the original GeoCities internet if people even know what that is anymore. When I was in high school and I really found I was good at technology. I grew up with computers.

[00:01:33] My dad's a physician and we always had the latest computer at the time, which was still like 16 color screen and stuff. But I grew up with using MS-DOS and Windows and things like that. So I grew up with technology.

[00:01:46] That certainly impacted that I was probably just generally good with technology. And then when the internet happened, it totally exploded my interest in communicating with other people. I also even found web design to be like a little bit of a creative outlet for me.

[00:02:03] I can't draw a stick figure to save my life, but I thought it was really cool that I could make web pages. So I taught myself web design and web development. that a little bit in college as well.

[00:02:13] But even in college in 1998, there were no courses in web design. There were like online tutorials that I would read. And then in medical school, I was like, I'd help fix people's printers. And I made like a little website to help us all share our contact information.

[00:02:31] That was back in 2003. And there was no Facebook. People had very basic flip phones. And we were a small class of 86 people. And so we all wanted to get people's phone numbers and get people's contact information and our AOL instant messenger names and our ICQ screen names.

[00:02:49] And so I just continued to make little projects and tinker like I still do today. And that's how I've always been interested in technology, even since I was a little kid, thanks to certainly my dad's influence on me.

[00:03:00] And so I think that's made technology come easily and naturally for me. And then combining that with medicine, I realized it was a unique skill set. I thought every doctor was good at tech. And that is that has turned out to not be the case.

[00:03:14] That's really great background. And I do remember from medical school, really thinking about you as one of the more technologically savvy members of our medical school community. I don't think that it was universal that the other future doctors in our cohort at Stanford

[00:03:32] really knew their way around using computers to do anything more than very simple things like word processing and looking things up on the internet. Would love to hear how you became somewhat of an entrepreneur by taking some of the dabbling

[00:03:48] that you've done and just helping others and figuring out ways to be more efficient using technology tools to creating MDCalc. How did that come about? I'm always just trying to solve my own problems, honestly, completely. I'm not trying to solve anybody else's problems.

[00:04:05] But my assumption is, has always been if I solve a problem for myself, the world is big enough that there's probably other people that have that same problem. And if I can solve it for me in a way that I like in a very self-serving way, that probably

[00:04:20] it's going to be helpful for other people as well. That's always how I've done things is just make things the way I want them. And then hopefully that means that other people will like them too.

[00:04:30] I'm very detail-oriented about wanting things in a particular way, and I'm very opinionated about how I want my stuff to work. And so really when I created MDCalc, I was solving my own problem. At the time, Jamie, you remember we would carry around these little like maybe quarter-sized

[00:04:50] books in our pockets and they had a plastic cover and little tiny little three ring binders and they would have like literally all of hospital medicine inside this little book. And there was probably one for peds and there was one for adult medicine, there was one

[00:05:05] for surgery and OB. And you would carry this around in your deep white coat pockets and they have literally probably I don't know what five point print that only young people could read.

[00:05:17] And they told you like the pearls about how to read an EKG, how to interpret somebody's low sodium levels, how to evaluate people for pancreatitis, all these things. And some of those things included some scores.

[00:05:28] There were a handful of scores back in the day, and these were scores that kind of predicted good or bad outcomes of patients. And people would ask you, oh, what would the score be on this patient? And I was like, why are you asking me to memorize this?

[00:05:43] I really felt like there's definitely stuff that needs to be memorized in medical school. And then as a tech guy, I definitely felt like there's definitely stuff that like, why do I need to use neurons for this when there's literally the world's information on a desktop?

[00:05:59] There wasn't an iPhone yet, so there really wasn't internet browsing on cell phones. But there were desktops everywhere and the world's information was available. So why am I wasting brain cells on this?

[00:06:08] So I literally just again, to solve my own problem, I was like, there were a couple sites that had a couple scores. And I thought, why doesn't anybody have all the scores? And I was also annoyed that you had to click a submit button.

[00:06:20] I wanted to like have one less click. And so I was like, screw it, I'll just build this myself. I know enough about web design and JavaScript and CSS. And I liked it a little challenge.

[00:06:32] And again, I built it the way I wanted it to solve my problem of, hey, I want the least amount of effort to find the score. I want it to search quickly. I want it to find everything. And MDCalc was born.

[00:06:44] I think I launched it with maybe 20 scores. And then over the course of the next whatever, three or four months, I just started looking for any score that existed to try to add it to MDCalc and probably had, I don't know,

[00:06:56] 50 or 60 scores by the time I got to residency. And so did you do anything to broadcast this or to help make others aware that this tool was available? Because I remember using it back when I was in medical school, but that was also because I knew you personally.

[00:07:11] And I think you let some of the other medical school folks know that this is a resource. If anyone wants to use it, feel free to use it. And we all definitely appreciated having that because back then I do remember those little

[00:07:25] paper books that were like steepled together. And I think they were, it was like passed on from the wise fourth year medical students down to the second and first year medical students.

[00:07:37] As you went up in your medical school career to your fourth year, you might end up kind of making modifications or adding to that. So it was the collective wisdom of the group, but having something available online was

[00:07:50] just so helpful such that you could be at any computer anywhere in the hospital. You just log in, you go to a URL and then you could quickly remember, okay, how do I calculate this BISAP score for, or anything else, whether it's dealing with hyponatremia or some of

[00:08:08] the other common things that you encounter on the hospital boards. Yeah. Those old hand-me-down pieces of paper, they're like the precursor to Google Docs, right? It was like a shared document that people like pass down to each other.

[00:08:23] And no, I probably the most advertising we've done with MDCalc is, like you said, I probably emailed all the med school classes and said, Hey, I made this thing. Give me feedback.

[00:08:32] Really, the only other trick I had up my sleeve was right when I made it in 2005, I still to this day read web design blogs for fun. I'm just like nerdy like that. I like seeing what's coming with the latest web browsers.

[00:08:48] But at the time, Google was really starting to take off in a more serious way and Google was starting to come up with ways that they would rank information. Now it's this thing called SEO or search engine optimization.

[00:09:02] And I had been reading about SEO for the past, I don't know, six months or a year or something like that. Probably even longer, honestly. And the SEO is essentially were some tips on ways to have ways how you should code your

[00:09:14] website so that Google can read it easily. And if Google can read it easily, Google will rank you higher. And this was I was blogging through med school. So this probably was a couple of years earlier than that.

[00:09:25] So I just designed MDCalc with good SEO, good search engine optimization. And from that was probably the other secret sauce that made it popular because if you search for and this is still true today, if you search for almost any score in medicine,

[00:09:40] we're the number one Google hit for that. That's how a lot of people find our stuff. Some people go directly to mdcalc.com or open up the app on their phones.

[00:09:49] Still a lot of people just type it into Google and find then just click on MDCalc from there. It really was just SEO. I can confirm that because just this morning, I tested it out myself with that BICEP score for pancreatic. Yeah.

[00:10:02] Now does come up as the number one hit on Google. You certainly did learn from those coding books back in the day. Yeah. How to read the website. So that's great. Really exciting to hear how you took something that was a pet project to help you better care

[00:10:19] for patients and have the data that you needed at your fingertips and made something that has been such a valuable resource for so many students, trainees, other practitioners in medical fields, both in the US and I'm sure internationally.

[00:10:36] We're probably used by at least two thirds of US physicians every month. And then certainly we don't have nearly as much information on the world population. But yeah, millions of people are using us. And because it's civic to clinical medicine, probably every user is probably a clinician, right?

[00:10:54] A doctor, a pharmacist, a nurse practitioner, because no patient is going to go on, go find the BICEP score or the Wells criteria for PE. It's so niche that we really are probably 100% healthcare professionals.

[00:11:09] And then, Jimmy, I have to give a shout out to my co-founder, Jo Haboo. Jo's an emergency physician as well. We met in residency and Jo certainly is the business mind that has helped to scale and

[00:11:21] grow and make MD Calc sustainable to keep it up and running all these years. We're going to celebrate 20 years in 2025. And it's certainly a large, it's been a really incredible partnership between Jo and I, between the tech and the content and the business mind.

[00:11:38] It's definitely the partnership that's made it go from like me coding a little website tool app and website and that's used across the world. Congratulations on being, well, it's that 20th anniversary. It's crazy. Yeah.

[00:11:53] And I think the other question is after the success of MD Calc, there's a parallel path that you could have taken where you just said, I really am good at this tech world. Obviously there was a lot of opportunity in Silicon Valley.

[00:12:08] You were coming out of Stanford, right? You could have chosen to go down a different path and not been a practitioner of medicine, right? But to just be a startup founder or somebody working in one of the many tech companies

[00:12:20] in Silicon Valley, but you chose to become an emergency department physician. You're still working as an ED doctor in the Kaiser hospital in San Francisco. Take us down that journey. And what was it that has continued to keep you grounded in clinical medicine?

[00:12:37] Yeah, I'm literally going into a night shift. I start nights to nights. That's why we're talking so early in the morning. So I could take a nap later today. Yeah, the logic I used to use this actually deciding about did I want to go to medical

[00:12:49] school too was could I do the others? What can I do as a hobby or as a side hustle or something for fun and what couldn't I? And I always came back to the fact that I can't do medicine as a side gig.

[00:13:04] Maybe that's changing a little bit now with more people doing consulting and part time work and stuff like that. But at the time it was like, I definitely didn't feel like I could be a good doctor on the side and then like mostly do tech.

[00:13:16] Because part of me was like, oh, maybe I should get into this web design thing. But I decided that I really wanted to do medicine and that I could always like change my mind and go back and do tech or go back and do policy.

[00:13:29] I was a social policy major in undergrad, but I couldn't go back and do medicine or be probably harder to do. And so that has that kind of really informed me to say, I want to finish. I want to do a residency.

[00:13:41] I want to be a good doctor. I enjoy it. I say that now, ask me tomorrow after my night shift and I'm tired. But like medicine is, I guess I feel like they're complementary.

[00:13:51] Medicine is a really unique set of like challenges for my brain and my body and my soul and my emotions and then programming and tech stuff is a totally separate challenge that similarly like activates me and gets me excited, but in a totally different way.

[00:14:09] I feel like there's different neurons that fire in the two different areas. And so that's why I still practice. I think both areas complement each other, both in terms of like my brain chemistry, but also they inform each other really well.

[00:14:23] I don't think I'd have nearly as clear and strong opinions about AI and technology if I wasn't practicing medicine, if I wasn't seeing patients and thinking about how these technologies might impact my patients and myself and the nurses I work with and my colleagues and

[00:14:42] the specialists I consult and all that. So I've found it to be really important. I think it actually makes me a much stronger advocate for like good technology and I find them to really be really complementary, but it's certainly a challenge.

[00:14:56] I certainly could have taken another path, but for me, it's been a really good mix. Yeah. And I really identify with that because I'm also a practicing physician. I do internal medicine versus ED medicine like you, but still I feel very grounded in the practice that I maintain.

[00:15:16] And even though I spend a lot of my time in the health tech world, in my role as chief medical officer of Memorial, for me, I think of myself first as a doctor and you think

[00:15:26] it's important then for me to actually be caring for patients and to continuing to keep up with the growing body of medical knowledge and to understand what's going on in the hospital setting and outpatient settings and really to have the pulse of that community if I need

[00:15:43] to then be working in that space on the technology side as well. Another question for you is that recently you've really been somewhat vocal and... Somewhat? I've been really making your name for yourself in the AI space, right?

[00:16:00] In terms of being a thought leader for how can AI be used responsibly in health? And I'd love for you to tell our listeners a little bit about how did you get into that

[00:16:12] space and I'm quite taken by the work that you've done with the Physicians Charter for Responsible AI. If you could tell us a bit about the genesis of that work and how you see that fitting into where AI is going in healthcare. Yeah, thanks, Jamie.

[00:16:28] Yeah, if you want to check it out, you can go to Physicians Charter, physicians with an s plural, charter.ai and download the summary and the whole document if you want. I really blame this or thank this on COVID.

[00:16:40] I think COVID certainly made me and probably everybody reassess their priorities, what they wanted from their career and their life and everything. And it was a challenging time for everybody, certainly challenging for me in the ER.

[00:16:53] And so I had a lot of things that changed about my life during COVID. I looked at my priorities. I got married during COVID to my wonderful husband, and I just figured out what I wanted to do. I turned 40 during COVID. I wanted to figure out my priorities.

[00:17:06] And so like I had written a column about emergency medicine for 10 years, a monthly column. So I decided I'm getting tired of this. Most of my columns are starting to be negative because I was burning out. So I stopped that. I stopped being as day-to-day involved with MDCalc.

[00:17:21] I still work with the team all the time, but not on the kind of day-to-day operational level. And I left Twitter because, yeah, Elon brought a bunch of COVID misinformation people back on the platform and I didn't want to support that anymore.

[00:17:33] And so I was just taking, I don't know, six, nine months off trying to figure out what I wanted to do next. And then honestly, ChatGPT launched. It was like in November of what, 2022?

[00:17:45] And instantly, literally it launched and I made an account that same day and I started night shifts that same day and I was just obsessed. I was showing everybody at work, look at this crazy thing. Oh my God, this is insane.

[00:17:56] And it totally re-engaged me and it made me both excited and a little scared for the future as I think it did for everybody. And I immediately was like, oh, I have opinions on this. I want to tell people my opinions on this.

[00:18:09] I felt like I could see a little bit of a bunch of different futures and where things could go. And I felt like I wanted to try to shape those futures again, both for my patients, like

[00:18:19] my worry that this would, this could potentially harm patients, but that could also be really helpful for my colleagues, for the whole healthcare system. And yeah, so then I was looking for a place to talk and I went on LinkedIn because I already had an account there.

[00:18:33] It seemed to be more professional, less kind of misinformation and stuff because you have to use your real name. And I just started, and then it also re-engaged that part of me that really loves to write as well. And so that's been super exciting, super fun.

[00:18:47] And it's, that's certainly been why I got so excited was like Chachabitty plus my ability to like feel like I'm an expert in this area and tell people and have strong opinions and also just write. I love to write and get my thoughts out.

[00:19:01] It does something to my brain, like with my ADD, it really helps my brain get clarity if I have to put it out on paper or on text or whatever. Yeah, that's been super fun, super exciting. And that's why I do it.

[00:19:15] It's fun and I feel like I can contribute and I feel like I want to have some impact. You mentioned the Physicians' Charter. That was just an idea that we, I met a bunch of people from LinkedIn and we all just decided to make a collaborative Google Doc.

[00:19:29] It's all practicing physicians from mostly in the US but from around the world in numerous different specialties. And we just, we all saw Chachabitty and we thought, hey, what do we want the future to look like?

[00:19:41] We should tell people what we want the future to look like again for ourselves and for our patients and protecting the patient-physician bond and relationship. So yeah, we all just worked on a big Google Doc together.

[00:19:52] I edited it and then found a guy on Fiverr that could put it together in a nice pretty PDF in white paper for us and then made the website for it too. And we're, it's getting to be about the one year anniversary.

[00:20:06] So I was about to put a call out for people that want to contribute again this year. So if any of us want to contribute, just DM me on LinkedIn or email me and I'm happy to

[00:20:16] see we wanted to look back at the past year and see what's changed. Where do we think the Physicians' Charter should go in for 2024? So exciting. Yeah, no, thank you. And for those who want to check it out, it's physicianscharter.ai as Bram mentioned.

[00:20:30] I do wish we had more time. I think we'll have to have you back on the podcast. Yeah. I'm going to end the conversation because I have to say, I really do appreciate a lot

[00:20:38] of your commentary that you tend to post on LinkedIn just about what is going on with AI and healthcare and even sharing about different research studies or other news that really are very relevant to many of us who practice medicine and are interested in how to better

[00:20:59] medicine with the use of technology. But really thinking about what is the impact on us as providers? What is the impact on patients? How are we doing this in a way that is safe and that is ensuring we're getting the right outcomes that we want?

[00:21:17] And we're making sure that we're very thoughtful about how we are using these technologies. So with that, Graham, I'll have you back to have another conversation another day. But really thank you for giving us some time today. I've greatly enjoyed this conversation and thank you.

[00:21:35] Thanks so much for having me, Jamie. It's great to talk to you. Thanks for listening to the Memora Health Care Delivery Podcast. For more ideas on simplifying complex care for care teams and patients, visit memorahealth.com.