Simply Begin (ft. Seth Hain)

Thanks for tuning to Digital Voices podcast, where we chat digital transformation, challenges, and opportunities across healthcare and life sciences. And now, your host, Ed Marx.

Hey, Ed here. Welcome to Digital Voices, another episode. I think we're like in our 200s. Thank you so much for listening. We know you have a lot of choices and different great podcasts out there. I listen to a lot myself, and the fact that you've taken a few minutes to listen to Digital Voices, uh, is very gratifying. So thank you for doing that. It really is all about you, and if we can ever help by having a guest that you'd like to hear on the program, certainly drop us a note and we can look into that. So today we have Seth Hain from Epic. Seth, welcome to Digital Voices.

Oh, thank you, Ed. It's great to be here. Appreciate the invitation and opportunity.

No, so, so fun. Um, so you're the SVP of R&D, which I really want to get into. It's so, it's so important, so cool. You know, you have to have that in any company that's going to be thriving and helping customers meet their objectives. So really excited about that angle. And I think the first time we met, we were trying to triangulate on it, but I've worked with Epic for many, many years, and one of the great things working with Epic has been That whenever we needed assistance or insights, you all would come down or we would go up. And I think you came down perhaps when I was at the Cleveland Clinic and we would just sort of roadmap the future and what's happening in development, where are we headed? And that way we could align where we were headed with tech at, you know, whatever organization I might've been serving at the time and make sure that there was this alignment and it worked great. And everyone was always so accommodating and And I wanna get into that philosophy a little bit later today. But Seth, the most important question I think of the entire Digital Voices is what's on your playlist? So like when you like to chill, what kind of music do you like to listen to?

When I— it's funny that you said when I like to chill. I would— the first thing that immediately came to mind is both the variety of things on the playlist, but then I've been deeply into ambient music lately. Which is, if folks are not familiar, a very calming and slow-moving but deeply interesting and introspective style of music, particularly some of the stuff from Japan. But I mix in a fair amount of other things, Nancy Sinatra, others as well, to make sure there's a little spice in life.

Yeah, yeah. These boots were made for walking.

Summer Wine is my favorite, but yes, that one's excellent as well.

Yeah, I, I, I love her as well. I'm a big Frank Sinatra fan, so, uh, I, I'm all things Sinatra. So that was, uh, she, she, she put out some good music, no doubt. Um, what about your life message or passion or mantra? Is there words that you sort of live by that guide you?

So it's a— I don't tend to think I have one. But a phrase that seems to pop up regularly for me is just simply begin. I think it both provides an opportunity to show the importance of just getting down to doing the work and that once you get rolling, something will happen and brings with it a little bit of humility that you don't always know right where it's going to lead you in the end.

Yeah.

But it's important to start heading down the path.

I like that. That makes a lot of sense. Yeah, because a lot of people, they just get stuck like deer in the headlights and they don't move, you know? And sometimes you just need to take that first step. That's cool.

I think it's been particularly important the last year. It has been a wild time in technology and it's been important to sort of roll up the sleeves and get a sense of what's possible with it, although you don't always know where it's going to be in the next 3 to 5 years. So that's something I imagine we'll get into in more detail, but It's been particularly apt.

Yeah, yeah, because yeah, it's the same in tech, right? If you, if you just freeze and do nothing, you will be left behind and, and certainly won't be meeting the needs of your customers and things like that. So, so especially in your role, that's why, yeah, this would be a really fascinating conversation. So just to give everyone a glimpse, a little bit more insight into who you are, Seth, share with us anything you want, you know, on your personal professional journey that sort of takes us to current state?

Yeah, so my background before coming to Epic was in mathematics. I studied pure mathematics. It's actually what brought me to Madison originally, was graduate school up here. Madison has a way that once you get here of keeping you. It's a fantastic place. When I came to Epic, my initial focus was on the backend architectures of our software. And helping it scale out to meet the expectations and demands of the largest organizations. And through that, I learned both the opportunity and the potential of integration from a software perspective, and also the importance of a strong foundation, right? Making sure that for those critical moments from a healthcare perspective, the system was both obviously available, being key, But also highly responsive. And so spent a number of years working with our performance team and our database architecture and kind of at that core level. And then I pulled together the math background with that and started really building out our artificial intelligence or machine learning team, if you will. And that has been a constant from an R&D perspective for me, that kind of intersection of analytics and AI that is the foundation behind our— the use of those across our applications. And then over time, I spent time with a variety of applications, including just spending a significant stint throughout COVID with our ambulatory clinical suite. So basically everything you walk in for.

Yeah.

And provided a nice perspective Where you're spending time on both that kind of foundational layer and set of technologies, but then making sure that for good reason, you bring it back to the front lines and deeply appreciating and listening to folks that are using the software and the importance of that. So it's been a journey. I've been here 18 years now, and that opportunity to continue to grow in a variety of ways. Throughout has been, has been very valuable and meaningful to me.

Yeah, no, that's cool. So yeah, Madison does have a way of making people stick around. What do you, are you, do you take advantage of the lakes or are you a runner, kayaker, or, you know, what, or what sort of things do you do?

The biking here is fantastic. Long walks with the dog just in the neighborhood are also very enjoyable. It shouldn't, I think are sometimes underrated. As my kids have gotten older, we've been spending more time on the ski slopes around here, which are not that tall, but they're still quite enjoyable as a family of 4 to get out and spend an afternoon.

Yeah, that's cool. Yeah, it's a great culture and a great environment. So what struck me when I was doing a little bit of research in 2015 is it just shows just how progressive the organization is. You had cognitive computing and ML You know, already in, in sort of titles and concepts, you know, because now, of course, the rage— and we'll talk about AI, of course— the rage is AI, as if it's something brand new. But you all have actually been working on it for some time, uh, and especially understanding cognitive computing and machine learning.

No, it's— I mean, I, I think one could certainly note that there has been a progression of, let's just say, algorithmic techniques even at the moment that bring intelligence into the software. And in ways it parallels what you saw happening in computer science, right? When the idea of artificial intelligence came out in computer science, it started in this foundation of expert systems and how do you encode knowledge into the software, which nicely parallels the rules-based logic every organization uses today For med-allergy interactions, those sorts of things. And then it transitioned into this phase of sort of deep statistical analysis. And, you know, as you were referencing 2015, that was sort of the rage at that point in time, was this opportunity to, using deep statistical techniques, predict things like deterioration on the med-surg floors or the likelihood that an individual would not show up for an appointment, and then being able to help put folks in a position to intervene earlier and improve outcomes as a result of that. And so that was a lot of what we were focused on, and at the same time, making sure throughout that it surfaced in the software in a way that always led to a human in the loop with the physician making the ultimate decision. So yeah, it's been a— A lot of those techniques just carry forward throughout, regardless of the exact type of intelligence behind the scenes.

Yeah, no, it makes perfect sense. So I do wanna jump into AI, but before we get there, the R&D part is so interesting. I think a lot of people have a big interest in it as well. But how long— but I just had this question, you know, you came from mathematics and then into healthcare. How long do you think it took to understand healthcare? Because I think some people would be interested in this answer because it's not automatic. It's a complex industry.

I appreciate the way you asked the question, although I don't know that it's true that I truly understand it yet. I don't think that this is an industry that one can completely wrap their head around at any point. It is changing too rapidly.

Yeah.

I think that it takes well over a decade to kind of get deeply into an area, but the importance of listening and continuing to hear from folks— you brought up how we come out on site regularly, both to discuss development that we have underway and are thinking about and to get your feedback on, but then Every developer, including folks that have just come here for the very first year, go out and do something we call immersion trips, which gives the opportunity to be elbow to elbow with folks using the software and listen and hear from what their experiences are. And I think that that's just critical as sort of an ongoing path to understanding the industry and as it evolves.

Yeah, that makes a lot of sense. You know, I'm smiling because I was just talking with a CIO, uh, who you would know yesterday, and she was talking about just in her organization to understand how to get from one building to the other took 3 years. And I remember talking to, uh, one of my best friends, uh, Chris Ross up at Mayo, and, uh, I think he was at his 5th year at the time at Mayo, and I was like, hey, do you think you Do you understand Mayo Clinic? And he's like, no, uh, not yet. I mean, it just takes a while, 'cause like you said, it's so complex. Then we have the double whammy, or not in a negative sense, a very fortunate sense to serve in technology as well.

Yeah.

So you've got this rapid change between healthcare and tech. That's what makes our, our vocation so, um, gratifying is, is, uh, working within this change to, to help save people's lives and improve quality care and those sort of things that we're we both, uh, have a kinship with. So yeah, so you talked about how, from an R&D perspective, how you stay connected to customers. What, what are your most creative moments? Like, is it in the shower? Is it with a customer? Is it, you know, where do you get— because obviously R&D, a lot of creativity.

Um, folks here give me a hard time about it. It, it tends to be at about 4:00 AM in the morning. I like to start my day with a cup of coffee and a book and some of that music we were talking about a little bit earlier. And I think that you need to put in that work upfront, both on the technology side and on the healthcare side from an industry perspective to have those ideas percolating. But then you need a couple of quiet moments to start to dig in and have some of those ideas. Um, come forward and start to flesh them out. And so I, I really value those early hours.

Yeah, that, that's good advice too for anyone out there listening that maybe struggles with creativity. There's something in the power of quiet and having that time alone meditating or reading or the ambient music, however you do it, the cup of coffee, uh, can really provide some inspiration. So thanks for sharing that. Before— okay, last question, then we jump into AI. Uh, what's one thing that you're most proud of so far? So you've been, you know, at this for 18-plus years. You know, what's the, what's the thing that brings you the most joy?

It's really hard. There's, there's 2 things that immediately come to mind, so I'm gonna, I'm gonna give both. I, I can't help it. Um, the, the first one relates to something I said earlier. I think the kind of, uh, Technically speaking, the speed and scale that one can have an impact on is critical. We now have 100+ organizations starting to build on top of the generative AI work we've done over the last 12 months. And that opportunity to help folks quickly and in a challenging time in the industry, I think, is is really meaningful. The other one, now on a bit more personal level, I've had the opportunity just to work with some great folks here at Epic that continue to grow and lead new products and push us ahead in new and challenging ways, myself included. And I think that each of those have just been really great opportunities and something I've been very happy to be a part of.

Yeah, no, that's cool. Yeah, definitely fantastic, uh, culture, fantastic people. And this is by no way an infomercial or anything for Epic. I just say this out of, out of my heart, having worked with you all for, for, for many, many years. So at the UGM, you know, Sumit, you know, did a— had a great keynote talk and talked a lot about AI. A lot of publicity has come out of there since, so we want to Uh, talk about it. So what is there a general, like a point of view of Epic on GenAI? We'll use the word AI, but we'll include in that broader definition, you know, GenAI and, uh, LLM, things like that. But is there sort of like this point of view or philosophy you all have on it?

Yeah, I think speaking kind of to generative AI specifically, and just to ground the audience potentially. This is kind of a broad term used that includes things like the engine behind ChatGPT as an example. And as an example of how we're using it in the software today, it can generate, as an example, draft responses to help a clinician respond to a message that a patient might have sent. They then can use that, edit it, send it off as an example. But do so in the context of that message that the patient sent in their actual text.

Yeah.

And, you know, 2 things come to mind broadly in regards to it. First, this isn't a new technology, right? Language models, deep learning, neural networks, all of these different computer science techniques have been around for decades in some cases. And I think What we saw in the last 12 to 14 months was a threshold crossed where they became viable for enterprise usage. They started to be a tool that we could integrate into software in a manner that at scale would help solve problems. So with that recognition in mind, we started to look at what problems were they best suited to solve. The one that came immediately to mind was helping improve clinical efficiency. So, the in-basket example, or inbox example, is one case of that. I think the other thing, kind of internally, that we've talked about, and Sumit mentioned this at the user group meeting as well, is that technologies do tend to go through these hype cycles, right? Generative AI may be at the peak of it. Maybe we haven't quite found the peak, but it's certainly on its way up or near the top at the moment. And it will come down for a moment as well. But if you look at the arc over time, it is this sort of force multiplier that just continues up and to the right. And so, you have these perturbations in the middle that we need to be anticipating and working through. But we want to make sure that organizations have the full opportunity to take advantage of it over that longer trajectory, not simply what happens over the next couple of months.

Yeah, yeah, that's a good philosophy. So you already shared the example of the in-basket notes, and there were some studies already published by some of your clients that were quite, uh, provocative. And, you know, even on the empathy side, like responses of GenAI, more empathetic than a typical physician response. What are some other examples of GenAI or AI? Again, I'll just stick to generic AI that you're seeing used by your clients.

Yeah, certainly. So, we mentioned briefly, and I'll just sort of talk about AI generally. For a moment. We talked earlier about predictive analytics and some of these purpose-built algorithms. Those are in broad use across the community at this point. You know, we have 1,000+ models in use across 450 organizations at this point. And I think one of the exciting things is we have over 200 model— custom models that organizations built themselves that they've deployed into Workflow.

Mm-hmm. Cool.

As part of that. And those tend to be in the kind of acute, often in the acute space. So things like predicting risk of falls, deterioration, sepsis, those events where if you get an individual to the bedside earlier to assess the circumstances, and that is key here, that's what the models do. They help folks get into a position to more to assess the context sooner and to help that patient earlier on, helping avoid a bad outcome ultimately, potentially as a result of that. That is one common set. Another is around improving clinical— sorry, operational efficiency. So using those types of predictions to help maximize the utilization of ORs, for example. predicting flow through the hospital to make sure your nursing staffing levels are at the right level, that you understand where beds may be opening up, et cetera. As you move into that generative AI space, which I think is a truly distinct set of use cases from the predictive analytics ones, although they both tend to fall under that AI umbrella, some new opportunities emerge. 2 of the most common ones that folks are in— that we're hearing positive feedback on are one, this general pattern of being able to draft text. So historically, folks have needed to document appeals letters for denials, responses to patient messages, drafting notes as an example. And generative AI can help in those contexts, creating something that directly references the message from the patient, the ambient recording in the exam room, or the denial, as an example of the 3 I just gave, and then help the physician move through their documentation more efficiently. So that is one very common pattern we see. The second one is with the kind of interoperable network across health systems that now take— is now in place. There's a lot of information at the fingertips of a physician and a nurse. And these models are very good at summarizing that information, always having the full medical record available behind the scenes, but having folks have information to start with that quickly pulls it together in the context of why that patient is visiting. If they're— you definitely want a different summary. if you're triaging a patient in the ED than if you're seeing them after 5 years for a physical, as an example. And that context can be taken into account and generate different summaries as a result.

Yeah, I love those. What about anything, and Seth, I realize that maybe for competitive reasons, you can't give a very specific answer here, but what are 1 or 2 transformational AI use cases that you you would consider put in the revolutionary category, you know, that aren't here today but potentially can, can be here?

Um, I'll, I'll touch on a couple of things. I think, I think there's some sort of transformational things happening in the AI space that are going to make these use cases adapt really, really quickly, um, and change. I think the first one is there's just a lot of opportunity for more context. And that's sort of both a technical and a layman term, if you will, where when you're working with these AI models, you can only feed them so much information. And what's happening over time is you can feed them nearly a book-plus worth of information now, but that's only really come about in the last couple of months. And so with that additional context, the quality and the tailoring to the very specific situations is going to continue to improve. And I think as a result of that, the efficiency will continue to improve. The other thing that I think is gonna be really transformational, and we're starting to see this in some of the consumer products around AI already, is the use of multimodal models. What that means is that there's multiple types of input, not just text, but you might be including an image as an example. A very simple example of this is we had a beautiful flower growing in our garden at home. My wife didn't know what it was. We'd honestly gotten the— we thought we'd gotten a different plant from the neighbor and we didn't know what it was, was the situation. And Could snap a picture of it and it could tell us what that was. Now, there's apps to do that historically, but this is general purpose that folks are starting to see it happen. But if you start to think of not just images, but things like genomic sequences, molecular structures, recordings in the— voice recordings happening in the exam room, All starting to factor in together, increasing that context, if you will, there's a real opportunity for discovery that starts to happen.

Yeah.

And I'm pretty excited about what might emerge as models better understand that, and then are used to help put physicians in the best perspective to care for patients and to work together with patients. understanding those suggestions and ultimately making decisions together about it.

Yeah, and then on top of that, you throw in quantum or, you know, the computing that's— the computing power that's so important that's starting to really, you know, escalate and help the whole AI. Yeah, then you're really talking some major, major game changers. So Seth, I had the privilege of serving at amazing organizations that were pretty high profile And so I had more resources than maybe the average hospital. So for listeners out there that are, you know, working at hospitals that are, you know, aren't in that, you know, top 20 US News and World Report, where do they start? So, so I'm a CIO at a, you know, 200, 300-bed community hospital. Where, where do I start?

Uh, start with a problem you need to solve. Um, I think Some of the fits and starts that I see often come from when folks begin this as an AI project rather than a quality improvement project or a clinician happiness project or an increasing access challenge that they currently have that they need to address. And then AI or whatever the tool is, comes along for the ride with it. And I think that it ends up providing a foundation for both the team and the technology to make sure that it has the maximum impact. Because ultimately, it's more about the workflow, it's more about the operations, it's more about the people involved than exactly the AI algorithm behind the scenes. And I think that that gets it off on the right foot.

Yeah, simply begin. Going back to your that works. You know, I I serve on the board of a couple health systems, and one of them, you know, is let's just say the average. And and I I love the fact that the boards so the boards are asking the same questions, right? Because they hear see headlines. AI AI they don't know AI, but they they know enough. They're smart people. So they're asking a question. I love our CEO. So our CIO, the CIO there is very well prepared and the CEO is great to begin with. And so he has a list in his back, literal back pocket, you know, of 78 initiatives that are AI-enabled. But to your point, there were problems the organization had that they needed to solve. And it happened that AI was the right tool and a lot of it leveraging Epic. So I always I always like that. So I hate to move on from AI because it's so interesting, but I, I want to kind of come back to R&D and leadership because I know, again, there's some listeners out there that are really curious. R&D, because we don't talk about it much in healthcare, um, and it's so important, but I don't, I don't know, I just think back to the, my own podcast I listen to or whatever, and I don't, you don't hear a lot of R&D people. So, what is advice that you would have for someone who's interested in R&D? So, how do you get into it? How do you get involved with it?

I think that a key aspect of it is the rate at which technology changes. And I think this is true outside of folks that wanna get directly involved in software development as well, but it's particularly apt for those that wanna do R&D, you need to be prepared for rapid change. So, learn how to continue to learn through that. You're not just gonna program in one language your whole life. You're probably gonna program in a different language maybe every year now. And now we're programming in English, or whatever your chosen spoken language is.

Yeah.

So, things change rapidly, and it's gonna be important to have a foundation that you can rapidly learn those along with it. And I think, you know, personally, I found that math was a fantastic foundation, um, for that for myself. There's other ways to achieve that as well. Um, by no means is one major specific to it, but I think that that's a key.

How do you stay current? We talked about how rapidly healthcare is changing and we can't even keep up with that. Technology is changing. You take time out early in the morning, reflection, think, read, listen, drink some java. And, you know, what else can one do to sort of stay current and keep their eye on the future at the same time?

Yeah, I have— interestingly, I have been going back and reading some of the old computer science texts. I think that there's this interesting parallel between science fiction books and some of the very early physicists and computer science thinkers, and quite prophetic in regards to what might be coming. And it serves as an interesting foundation where then I pair that with research papers that are coming out immediately And provide a sense of direction. And there's an internal kind of chat that we have here across our research and development group, and everyone working on these projects is regularly adding in the latest research papers that they're seeing or test results all the way to— I always find it entertaining because Sumit puts his posts into that chain late in the evening with the papers he's looking at, and then mine come in at that 4:00 AM. And so there's always this flow throughout the day. So I think a combination of sort of looking back and looking forward is a good way to balance it.

Yeah, yeah, I like that approach. Yeah, because you have to always remember history and a lot of the early works and not lose sight of those as you as you look to the future? And what advice would you give new college grads, you know, maybe in general? So it doesn't have to be R&D specific, but like, you know, the university there, they— you were in their master's program. I predict someday the PhD program. You sound like a doctor, philosopher. And what would you say? You're talking to the graduating class.

This might sound contradictory. The first one is they've got a real opportunity as they come in with fresh eyes. One of the great things about working here is so many folks bring in new perspectives and on an ongoing basis, and the opportunity to learn from them and to be challenged by them is great. So always encouraging them to speak up and to share that, even And it provides a lot of opportunity for everybody. The contradictory part of it is, I think we all can listen more. And there is an opportunity, and in particular, that's why I think that those chances to get onsite elbow to elbow with the physicians alongside your leadership teams, as an example, Ed, and listen to both the challenges and the opportunities that they see. And then bring that back and think about what opportunities can open up when you combine them with the technology. So I think that that has to be the second piece.

Cool. Seth, this has been so informative. I really appreciate your time. We talked a lot about different things from music, what to do in Madison when you come to visit UGM, all the great things that you can do, the whole concept of Simply Begin. We talked a lot about AI, what Epic is doing with AI, and how just an average hospital can start moving along and taking advantage of all these tools and capabilities. We talked a little bit about leadership. Is there something I missed or is there something we should double down on? I'll give you the last word.

Simply looking forward to 2024. It's been a wild ride in 2023, opened up a lot of new opportunities. And looking forward to what the industry can do together in 2024. So appreciated you inviting me on, Ed. It's been a great conversation.

Well, thanks, Seth. And one last question before I let you go. I should have asked it at the top, but I didn't want it to cloud perspectives. Vikings or Packers?

Huskers. I grew up in Nebraska. There's only one answer.

Okay. How are they doing, by the way?

Oh, we don't need to bring that up. It's okay.

Okay. All right. Well, that wraps up Digital Voices and go Nebraska.

Thank you for listening to Digital Voices podcast with Ed Marx. If you enjoyed this episode, subscribe on your preferred streaming service and leave a rating and review. And most importantly, Thanks again for listening.