MGMA Business Solutions: Transforming Primary Care with Human-Centered AI
Download MP3Well, hi, everyone, and welcome to the MGMA Business Solutions Podcast. I'm your host, Daniel Williams. I'm a senior editor at MGMA and pretty much the host of the MGMA Podcast Network. So you've seen me on here before. Today, we are really excited.
Daniel Williams:We have a new guest here, Phil Tornroth. He is a vice president of engineering at Alation Health. Phil's been innovating in healthcare technology for over two decades and helping build tools that reduce clinician burden and bring human centered design to the forefront. What we're going do is get to know Phil a little bit better, get to know a little bit better about Alation Health, but also something that's exciting and on the top of mind for a lot of people, getting to understand what is going on with AI developments out there and how they can support primary care clinicians. With all of that said, Phil, welcome.
Phill Tornroth:Thank you, Daniel. Happy to be here.
Daniel Williams:Yeah, so good to have you here. We have a connection. You're in the San Diego area. I've got a daughter in college down in that area, so it's really good. Maybe I'll swing by and say hey to you sometime.
Daniel Williams:I never need an excuse to go out to San Diego. I love it there.
Phill Tornroth:Yeah. Me too. Weather's great here right now too.
Daniel Williams:Yeah. So let's start out, Phil, just getting to know a little bit about your background. I I shared some information, but what are some things that you might wanna highlight for our audience so they could just get to know who you are?
Phill Tornroth:I've made a career in health care startups. Those that those are basically the two through lines for for my career mostly is almost exclusively in healthcare and almost exclusively at relatively early stage venture startups. I did the sort of first half ish of my career, specifically focused on patient safety and preventing medication errors in hospitals. And I think that gave me like a good sense of why I wanted to work in healthcare. And I'm really grateful to have had such a mission driven start.
Phill Tornroth:I think maybe if I had started in like hospital billing or something, I could have run for the hills. And then for the last fifteen years, I've been focused mostly on independent primary care with Alation.
Daniel Williams:Okay. I have to ask you. We're gonna get more into AI, but is anything going on in AI surprise you? Or do were you enough into that field where you went, this is the direction it's going? Did you have any inkling?
Phill Tornroth:No. It surprised me. The the truth is I so my my real AI awakening in terms of what we're gonna be able to do practically in the relatively immediate term, I think, came, like mid twenty twenty three.
Daniel Williams:Okay.
Phill Tornroth:And before then, I you know, I was paying some attention to machine learning, but I didn't see us rounding the corner as fast as we have. And in truth, I've always been like a little bit of a of a Luddite when it comes to new technology. I like to see it work before, you know, I get too excited about it.
Daniel Williams:Yeah. That is so cool to hear from someone who has a technology background that you're not just right there at the forefront with everything, every new gadget and, iteration that comes out. So that is I'm not the only one. Let's learn more about Alation Health. I know that you guys are a partner of MGMA, and we're great to so glad to have y'all on board with us.
Daniel Williams:But what I'm seeing here is that Alation Health supports over 36,000 clinicians and is known as the largest community of primary care innovators. Tell us a little bit about Alation Health. What do our listeners need to know? Our listeners are those medical practice leaders and decision makers.
Phill Tornroth:Well, first of all, founding story is pretty powerful and had a lot to do with me joining. So Kina and Conan were our founders. And they have a father who's a practicing primary care physician who just retired from his practice but worked for many years in Walnut Creek. This was around the time that like Meaningful Use was showing up and there was an enormous amount of incentivizing providers to get off of paper and onto these EHR systems. And they helped their dad survey the field of what was available and came to the conclusion that these systems were going to be really, frankly, existentially terrible for these, especially these small primary care physicians like their father who provided really complex proactive care to a panel of 1,200, 1,300 patients.
Phill Tornroth:And the company was founded on the belief that primary care is our biggest lever against downstream illness and costs, and that it's frankly, like undervalued as an opportunity. And that attention would turn in that direction and that technology was not on track to help provide leverage. So we started Alation to focus very specifically on that.
Daniel Williams:Okay. Give us an idea then. What time frame are we talking about? When was that?
Phill Tornroth:I think they started working on it 02/2009, and then I joined as the first engineer in 2010.
Daniel Williams:Okay. All right. I wanted to follow-up then about that work with primary care. So how does this community of Alation Health supporters, supporting over 36,000 clinicians, how does this community influence your approach to integrating AI tools like Note Assist into your EHR platform?
Phill Tornroth:Yeah. I mean, it was one of the things that we would talk about a lot is having a clinical first strategy.
Daniel Williams:Okay.
Phill Tornroth:So to like sort of double click into what I was just saying, I think when Keenan Conan looked at why technology wasn't providing as much leverage to clinicians as it ought to, and as it did in plenty of other verticals where technology was really reshaping workflows, one of the things that they noticed is that virtually all EHR companies are billing companies at heart, and then maybe eventually get around to building an EHR or buying an EHR, but their focus is really on billing. And we decided first and foremost to take a clinical first approach and focus first on what do providers need in order to make this really powerful diagnostic tool, because that's what the chart is work for them. And how can we add leverage? And when it comes to AI, a lot of our thinking is centered in that same place where the first place we go is what clinical workflows look like, and in what ways could this technology reshape and enhance those. And sometimes that also means investing in billing.
Phill Tornroth:We have a billing platform and an all in one solution today, but it mostly centers us and what do providers start the practice to do? And how can we make that as joyful and efficient as possible? And how can we make as much of the other stuff go away?
Daniel Williams:Okay, thanks so much for that Phil. Now, in reading about Alation, the EHR platform is known for its intuitive design and the way that has unified work flows. So when we think about it that way, how are you designing the products? How are you designing things that can be implemented for those medical practice administrators and other people in healthcare?
Phill Tornroth:Yeah, when we talk about unified workflows, a big observation we have when we look at systems and features that are failing in healthcare is that they often feel compartmentalized. I think in some cases, the teams working on these might shift their org chart a little bit. So, you know, really one of the key innovations in Elation early on was just making it possible to see disparate pieces of data at the same time to be able to look at a patient's medication history next to your last visit note or review historical records while referencing your own. So when we talk about unified workflows, a lot of it is thinking about what the purpose of this is first, and then designing surfaces that support that clinical workflow. We made the decision last year to really lean in on AI natively.
Phill Tornroth:And so by that, I mean, we hire and train and build these skills in house versus depending entirely on partners to build solutions on top of our system. And one of the reasons is that we really firmly believe we won't be able to accomplish those unified workflows in a really graceful way and incorporate AI the way it should be incorporated without doing that. If we depend entirely on partners, we're going to get little pieces of the puzzle that don't really connect to each other.
Daniel Williams:Right. I noticed that earlier I I mentioned the term note assist, and I didn't get you to elaborate or define that. And so we're just talking about it like everybody listening. Oh, yes. Of course, Noticist.
Daniel Williams:What is Noticist first? And then I have a follow-up for one of your new innovations in that program.
Phill Tornroth:Yeah. So Noticist was our our flagship AI product that we launched last year. It's a natively built AI scribe integrated directly And into so that means that it records and transcribes your conversation with the patient. And in real time, every few minutes or so, it transforms that into note content. It respects your visit note templates, and it'll fill them out for you.
Phill Tornroth:It respects the problem lists and incorporates context that's already in your chart. So it because it's natively in the application, it has all that context and you're not cutting and pasting and assembling this information from another tool.
Daniel Williams:Okay, that is so helpful. So my follow-up is a note I had from you is one of your latest innovations is something called Actions. I feel like I'm talking to, you know, somebody from Apple or something here. It's all these really cool iterations and new products. I'm just I'm so excited about it.
Daniel Williams:I love this stuff. What is actions? What is that?
Phill Tornroth:Yeah. Actions was a logical next big step for us. Know, one of the great things about Noticist is that you can focus on your patient. And we have a lot of doctors that tell us like, this is amazing. I don't have to keep jotting things down and using this half of my brain to reshape information.
Phill Tornroth:I can stay focused on my patient. But one of the things that we still saw providers doing is needing to either like review their note and make sure that they caught every to do item that they needed to follow through on, or some of our providers would actually like have gone back to a piece of paper and we're keeping like a checklist in front of them. And so what actions does is it listens to all of the things that you intend to need to follow-up on in that visit. So if you talked about refilling a prescription or changing a dosage or sending a lab order, and also the things that might not be as top of mind, like if early in the visit, you'd mentioned that you were going to get the patient like a handout on a certain set of exercises or whatever. So we keep all of that in the to do list.
Phill Tornroth:And at the end of your visit, you've got it there. If you take actions in the application, so for instance, if you go to our EHR and you write that prescription, we notice that you write it and we check it off for you. And so it's this live list of the things that you intended to follow-up on.
Daniel Williams:Okay. This may be a simplistic question, but I I need to ask it for myself and hopefully for some of our listeners here. When I work with AI and prompt and do that those sorts of things, sometimes I'm not getting the results I want. So then but I know what I'm looking for. So I go, no.
Daniel Williams:That's not exactly what I'm looking for. Let's shift it here, here, and here. Now you're describing when y'all are building out your program, your AI, you're doing that natively. So how do you ensure that it's gathering the information that you need? That's something that y'all are building into it on the front end, or is it something where someone comes back like a medical assistant or someone just to make sure that's all being integrated, all those notes are being done the right way?
Phill Tornroth:Yeah, so our tools right now, the ones that we've shipped are pretty agentic. So the AI listens, records, remembers what you do. In terms of making sure that's accurate, I think there's two big levers that we talk about and use as a team. One is we do quite a bit of pre formal evaluation. So we build corpuses of data and we run through enormous numbers of test cases and we're constantly iterating and trying as new models come out and things.
Phill Tornroth:That's really important to us. And it's actually most of the time and effort that our AI teams spend on these features is writing and running those evals. The other lever that we use is design. So building with AI is really like designing for failure and iteration and how can we make like one of the sort of jokes that's become a mantra on the team is this there's this Mitch Hedberg joke about how escalators don't break, they just become stairs. And so we talk about that a lot on the team is like, how can we design an experience where we make failure safer?
Phill Tornroth:Because these tools are gonna get it wrong sometimes. And so can we design a surface and an experience that allows the provider to still be in control, see what happened, recover from these situations?
Daniel Williams:Okay. Love that. Thank you for clarifying that. So let's move to the next step here. Technology, as we know, can help so much with workflows, but can also if somebody's not tech savvy, some of our healthcare professionals aren't just up on the latest, greatest technology that's out there available to them, we don't want to disrupt workflow.
Daniel Williams:So how do y'all integrate things? How do you design it so that it doesn't actually impede the efficiency in the practice?
Phill Tornroth:Yeah, absolutely. Mean, we do think about that quite a bit and making sure that these experiences have, like we'll talk sometimes about these experiences having off ramps or on ramps and like our, since we're all learning how this technology is going be integrated into our lives, and there's such a variance in fluency, we think a lot about making sure that these are experiences that you can dip in and out of or that we have ways for you to learn about them safely. Yeah, also, you know, the point about tech savviness, like I mentioned, I'm kind of a bit of a Luddite and come late to these things. One of the things that I found so inspiring about AI specifically is that that adoption curve is kind of reversed. Really early on when we started working on Notisys with some of our customers, I talked to a couple customers who ran clinical teams, and they were finding that the biggest adopters of these tools, the people who latched onto them the fastest were the ones that had the hardest time using a computer.
Daniel Williams:Wow.
Phill Tornroth:Makes sense. These are providers that have been on paper their whole life. And now the idea that they can talk to their patient as opposed to having to click and type through this note, It was just a much more human experience for them. So I've continued to see that even on our own teams, and I love that about this technology in particular.
Daniel Williams:Yeah. And that that leads into the next question I have for you. I have a note down here that you describe yourself as a reluctant technologist and really with your focus on human centered design. So let's talk about that, you being a reluctant technologist, and then how that goes into the products that you build out.
Phill Tornroth:Yeah. What it means is I I've never really been enamored by technology for technology's sake. I've always been much more of a product and design person who happens to do software engineering as a trade. And so don't worry, I hire and employ lots of people who are fascinated with technology and they're the ones that push me to look at things and adopt. I'm mostly the the the sort of like, person that, like, puts the brakes on and asks, like, can we show value with this yet?
Phill Tornroth:Like, is this is this gonna make a real impact on real people and push us to experience that?
Daniel Williams:Yeah. I've got a couple more questions for you before we sign off. So as we all know, I have conversations almost daily with peers, with everybody I interact with. I had this conversation recently with someone. I've only had a smartphone for about fourteen years, but I'm not young, and I don't remember life before having a smartphone.
Daniel Williams:I was telling you offline that I was just in Spain, and so I didn't have a a SIM card. I wasn't connected, you know, where I was. So I actually had a detox of the smartphone, the device, everything because I wasn't staring at it every five seconds. You know? I've adopted back.
Daniel Williams:I'm back to staring at it. But what I wanted to get to was with AI, as you were saying, your adoption was around '22, '23 of you know, just a few years ago. And we're already like, well, what would I do without AI in front of me? And it is evolving so rapidly. Where do you see it going, you know, in just the near future?
Daniel Williams:Where are you looking around the corner? What's your team doing as far as where you think it can really make seismic shifts in health care in a good way?
Phill Tornroth:Right. It has been so fast. And there's so much opportunity that I sympathize. Like, I think a lot of us just sort of want to, like, hit pause and breathe and try to figure this out together. It's super exciting.
Phill Tornroth:And it's super exhausting all at once. One of the things that we've been looking for is especially now that we have, you know, we went through this process of building an AI team at Alation, which I think a lot of companies did and build their expertise in one team and that team built and shipped note assist and shipped actions. And then really over the last like six, nine months, we've made all our teams AI teams. So we've made this investment in building up the fluency in the entire team. And so that means going from student to teacher really fast for our experts who have only been AI experts for a few years in this new world.
Phill Tornroth:And so, we've been giving them a lot of guidance. One of the things that we've been saying quite a bit is, you know, there's this confusion with AI because, you know, on one hand, the the sort of hype cycle which have have you telling us that super intelligence is around the corner and this stuff is genius. And on the other hand, you know, you've got this sort of alien technology that can pass the bar but struggles with arithmetic sometimes. It's It's really hard to reckon those two things. A lot of the opportunities are just, you know, this is technology that gives computers common sense.
Phill Tornroth:And yeah, it doesn't have to be genius, but you can do a lot with computers that have common sense. And there's so many edges that we can round off of these systems that I think we're all a little blind to because we just take the sort of like lack of flexibility that computers bring for granted. And so even in elation, we're kind of rewiring our own brain and sort of like reminding ourselves nobody wants features. Nobody asked us to build features. They're not virtuous.
Phill Tornroth:People asked us to get their jobs done and have joy in their work and to be effective in what they're trying to do. And so we're sort of going back to first principles on a lot of these workflows and saying how much of this is here because computers didn't have common sense. And if we reimagine these surfaces completely AI native, let's try to pretend, for instance, that we were born with that smartphone in our hand, like, what would this look like?
Daniel Williams:Yeah. Last thank you for answering that. Last question. I see over your shoulder, there's an emblem there, Elation. I have some notes here.
Daniel Williams:It was recently awarded best in class for small practice ambulatory EHR PM. At MGMA, we love celebrating along with our partners, with our MGMA members. Talk about that award and what went into it and, you know, how you're kinda riding that wave right now.
Phill Tornroth:Yeah. We're super we're super grateful to to have that, to have gotten that award. The team's really jazzed about it. Yeah, I think it reflects a lot of the hard work that we've done thinking about the experience. And one of the things that I love about the class surveys and those awards is that they're out there talking to our actual users.
Phill Tornroth:And when we get these reports back, get quotes and we get feedback. And they're so consistent with the feedback that we solicit from our users on a daily basis that it's just it's it's great. I've been in other companies and areas where you wonder about some of these awards and how much is marketing fluff and does it really reflect the experience that actual users are having? And I've been so happy for the the class institution for that reason.
Daniel Williams:Yeah. Well, Phil Tornroth from Alation Health, thank you so much for joining us today on the podcast.
Phill Tornroth:Thank you, Daniel. Yeah, I really enjoyed it. Would love to be back sometime.
Daniel Williams:Yeah. You may see me show up in San Diego. Can go get I'll a cup of
Phill Tornroth:be here.
Daniel Williams:For all of our listeners, what we're going to do is we're going to put some links and resources to the things that Allation Health are doing out there in healthcare and how they can provide solutions for you at your medical practice. Until then, thank you all so much for being MGMA podcast listeners.
