What a Software Developer Learned Running a Cancer Center

Doug Flora: Welcome everybody. We are back with episode two of our new series where we're trying to break down the barriers to adoption, introduce you guys to these crazy concepts around AI, digital health and oncology. This one's more about infrastructure. This one is about architecture and about discipline. And I think that's maybe not the sexiest topic, which is why Sanjay loves it.

And he's really going to get into this, you guys. It's going to be fun. And our guest is an expert at this, probably one of the best in the world at what he does. And we'll explain why. And bear with us on the tech stuff because you will be glad you did because these questions are going to come up in your office.

They're going to come up in your boardroom. They're going come up with your COO, your CEO, your CMO, your CMIO, all your O's. So with that, I'll introduce the team. You guys know Sanjay Juneja, the Onc Doc. He is ubiquitous and effervescent as the teacher of oncology AI across the country, probably the most followed oncologist in the world for this topic.

We're glad to have him. And over here, we've got Clint Taylor is our industry executive leader. But Clint is a serial entrepreneur, has lived in tech his entire life. You'll learn over the course of the series, we really wanted him here to let us know on the oncology leadership team, of their problems too. He's become a close partner to a lot of us in the industry.

So, and he's had great success. And our star for today is my friend, Jeff. Jeff Hunnicutt, you probably know well. He's one of the national thought leaders in just about most of the things. When I was coming up, it was ACCC.

Co aboard now, he runs one of the most innovative, most thoughtful practices in the country at Highlands. And we'll talk about why they're succeeding and it's not an accident and it's not luck. But the story is that Jeff is not a physician leader. Jeff came out of this world and how fortuitous for Highlands that they have a guy who trained as a computer engineer before, you know, that was important.

Sanjay Juneja: Yeah. I am super excited about this episode and Jeff, I'm just going to tell you where we're going to get right into it. I have an hour with you. We have an hour with you. And Byron, you're just very, you've had a lot of experience the last two years of having a ton of people come to you and you've had things that are point solutions that we'll talk about and platforms, and you have to take into consideration so many of the different pieces that are required.

Are you asking about LLMs? Are you asking about how often they're validating things? Tell us your workflow on just really kind of even on the surface understanding the durability of a company.

Jeff Hunnicutt: So it's a well thought out question. The answer is yes for all of it. But to dive in, think the more important question is probably, I guess, what are the fundamentals for how you're approaching it? Maybe first, What are the things you're looking at? The vendor space, as far as new technology solutions, is really crowded.

I think practices are getting inundated with these different contact points, whether it's through LinkedIn or it's an email or some other digital method of trying to get time with a practice leader. You have so many different people that are entering the market. So for me first, when I'm sitting down, yes, I want to understand what's the core technology offering? What are you bringing to market? With a lot of these startups, it's usually a single solution at this point.

What we're more interested in is what is your appetite and your aptitude for doing a host of adjacent tools in the same space? So essentially, is this a slide deck or is it a product? Is this a product or is this a suite of products? You need to be able to answer those questions, I believe, to be able to understand if this is a partner that you want to link arms with for the next five years, because you guys know as well as I am, I've never seen tech innovate so quickly as we have in the last three years. As that's the case, going with a single solution partner, what you'll find is that you're probably going to have something else that's to swallow it up before you know it.

So for me, it's the thoughtfulness that goes into that first few steps of determining can we actually work with these people for the long haul? Can they be a truly be a partner, not be vendry? To quote Doug. I like

Sanjay Juneja: to make up my words.

Doug Flora: I like

Jeff Hunnicutt: that by the way. Use that.

Doug Flora: Strategery too. Use it a lot. Strategery. We have a lot of strategery.

Jeff Hunnicutt: But those are really, really important things. From there, I believe fully involving the team inside your practice, building a culture of change management before the change ever comes is a really important thing. So if you're involving different stakeholders from your team in the process of being that next vetting engine, so to speak, and you have people from all over the company, it could be an RCM product as an example that you're looking at, but involve nurses in the process. May not impact them at all, but they're witness to the change. They're actually able to see, Oh, well, these guys are doing something.

So by the time it comes around to them, well, maybe they're more bought into it. From there, yes, sure, it hands off to network compliance and security and all the things you have to check, but those first boxes are probably more important.

Sanjay Juneja: Yeah, I mean, I think it's, Emily had said the same thing in our first episode is really getting everyone involved. You really underscored something that Deborah Pat also says is really having the change management kind of like the fidelity of the institution prepared to have a viable solution. One of my worries is that people consider getting a point solution as kind of their way of breaking into the space, but it's really not that, right? I think people don't want to make this big platform leap or think about all the moving parts, but unfortunately, even a simple solution does require so many different things. What are some of those things that you would recommend thinking about when it comes to that preparation of the change?

Is it thinking about your internal security? Is it thinking about your infrastructure? Are you normalizing things and wanting to be in charge of your data elements? Are you using a vendor for that? How does that process look to even have the setting ripe, so to speak?

Jeff Hunnicutt: Big question. That was a lot with several different factors to the answer. I'm going zero in first on what you said regarding data. In most practices, data is living in all kinds of different siloed databases that don't touch each other. Maybe there's an HL7 interface that connects them, but nothing in a meaningful way to where you can really turn an apples and oranges equation into apples and apples.

In our practice, something that was very, very important was for us to get access to those different data sources, centralize them in one repository, normalize the data so that you can actually build your engine on top of the one source. So all of a sudden, I can more accurately compare or I can integrate ePro data into a discussion with my clinical outcomes and do it in an automated fashion. That's an example of a baseline infrastructural type of change that a lot of groups don't, they don't think of that part. But if you've done that, and then you have the vendor partner relationship that requires access to different data elements, well, the infrastructure is already there. You can do more, you could be more effective.

Become a kid in a candy store of the possibilities of what you could do in that environment.

Doug Flora: This is what you guys do best. I love that you guys didn't bury the lead because I wanted to get to this too. This is where Highlands wins, right? And Northwest Arkansas is not typically thought of as a bastion of healthcare. No offense, neither is Northern Kentucky.

Right? It's not Boston, it's not Houston, but they're ahead. And so I think what you guys have done well, and this is for Clint, probably because you spent your entire life in tools. You can't automate what you don't have organized, right? If you don't understand that first fundamental step, everything else is going to be a disaster because the tools don't talk.

And so that's, I'm really glad you brought that up. For us at this table, AI is not a product that you install and you've got to get away from thinking about that. It is an unglamorous heavy lift to make sure you're organized, that you have infrastructure, that you understand your own data governance, that you have the right people at the table. Because the clinicians are coming to you with point solutions. I need this to look for lung nodules.

I'm a pulmonologist. Well, that's great, but maybe I can buy a program that does lung nodules, clinical trial screening, navigation, etcetera, with one vendor contract and others. For you, you're the designer. Your team has been doing this for a long time before vendors show up. Maybe the question we all have to understand is how are we organizing our lives and what problems are we trying to solve so that we

Clynt Taylor: can help you guys figure those things out too. You know, recently I've had the opportunity to work with a big software vendor called Candida. And one of the big things they always try to do is go in and help companies get ready for the vision they have for using AI. And I learned a lot in that process. I mean, the years I spent there really gave me an appreciation for how much work that is.

It's not an easy thing necessarily to do what you were just describing. And it's an effort, it's a focused effort. And you have to have a vision in why are we doing it?' has to be a real reason for the why to do it. Not everybody can start there, but not everybody has the infrastructure or the resources to do what you did, so smaller practices might say, we need to find some help to do that. But it's super important, you said, Doug, to do that regardless of where you start, being ready for that is a huge, huge part of it.

Doug Flora: Yeah. Think it flows very well. Jeff, you may not have seen this, better soon because you're going like it. But Emily's episode, we talked about how do you get started? And I think probably the narrative here is, she set the power of the possible and the people who are listening to this conversation, unfortunately, are going to have to learn what it costs too.

Right. And what it costs is discipline. And talk a little bit about your leadership team. We know Lacan quite well. He's one of our dear friends as well and brilliant guy who also understands the big picture of sustainability and investment in things that later may pay off, but are a little bit hard early on.

And so how do you guys manage the change management? You've got 60 docs, is that right?

Jeff Hunnicutt: 30 docs, 30 advanced practitioners.

Doug Flora: Okay. That's 60 providers. Yes. Thanks. And we talked a lot in the first episode about change management.

How do you, as the engineer, talking to clinicians, and I say it's a feature, not a bug that you're there. So how do you tackle that as their CEO and lead them to the promised land?

Jeff Hunnicutt: Yeah, referenced this earlier, but building a culture of change before the change comes, I think is the way I described it. To me, it's vital for us to be able to survive as we go forward as practices, independent oncology practices that many the different groups have the walls closing in. If we can't change, if we can't adapt, it's not a good future for the practices. What that looks like at Highlands, it's fairly simple and it's going to sound somewhat cliche, but the creation of a technology committee. Now what the tech committee actually does is probably the more innovative part.

As I mentioned before, it's that you have relevant stakeholders from across the practice. So you're evolving all these different people in that process. So every single tech tool that comes through, they're a part of not only seeing what it does, how much it's going to cost, what the ROI is, but they're a witness to IT being in there, talking about SOC two compliance and talking about data infrastructure and all these kinds of things. Compliance is in there as well, so they're talking about, is this going to affect anything with HIPAA or data sharing? Everybody is a part of the process.

So an RN on the floor is learning about these things that they never would have learned about otherwise, but they also feel like they're a part of the decision making process because when it comes down to the end, everybody is essentially voting on, do we do this thing or not? Does it meet all the standards that we have as a practice? Is it going to help us go in the right direction? So when you say yes and you put the rubber stamp on it, your buy in is very tangible. From there, you have to teach the team and to really instill it in them that the responsibility they have is to then take what they just learned, take what they disapproved, take what they just experienced and go to the rest of the team.

Now you perpetuate the message downstream. So now the infusion center knows that there's a new tool that's in imaging that's gonna speed up lung cancer nodule identification through AI. Wow, how cool is that for our patients? It wouldn't affect them in day to day life, but they know that something cool is happening in the practice. So when it comes to be their time and now you've got chart prep or something like that, that's affecting the RNs and the MAs, and they're feeling like they're a part of positive change too.

But you have to do it years in advance of when the tech actually comes if you want

Sanjay Juneja: it to be truly effective. That's such an interesting take. I didn't think of it in that manner. It's, know, a lot of people think, okay, inform when it's coming, but really inform something that's tangential just to even break in this concept of, of, you know, potentially catalyzing change. I'm curious, it sounds like obviously the team internally is just so key and you've highlighted that, but unfortunately, I guess there's a very variable range of how prepared perhaps a team is like a tech team.

I've had a few texts from friends that are colleagues and fellows, they're like, Sanjay, quick, they assign me on the tech board or the AI board for my hospital. I don't know anything. Where do I get started? Because I'm going have to participate in this process. Mean, some people are actually being thrown into this you know, by virtue of just being the youngest, you know, position in the group.

With that said, I, I worry that there's this very labile needle of how much is done at the institution level when you partner with, with a piece of technology and how much you rely on the company for. Meaning, the abstraction quality, the curation quality, the model performance, the validation, the auditing to make sure that once it's propped up, is it performing the same way it did at this other center because it's the first time in Arkansas? A lot of that hopefully, and I think Clint can speak to this is done by the vendor. And these are things I'm sure you've vet comprehensively. But I'm sure in your case, that needle isn't just full reliance on them, but you have a needle that, compared to most is also close to your facility instead of taking their word.

How do you set that up? What does the conversation look like as a partnership saying, okay, this is what you're going to do. This is what we'll do. And then do you kind of, in the background also double check what they were supposed to do?

Doug Flora: How does that look like?

Jeff Hunnicutt: Yes, you do set up those double checks to be able to prove, is this actually materializing the way that we set? Now there's only a certain amount of, or a certain level of insight that you have past the wall of I what the vendor can't go in and guarantee and see, are they really, really have the security measures in place that they said they did? There I can ask for reports. I can trust those things are accurate. I can't prove without a, with no shadow of a doubt that they are truly compliant.

There are situations that have happened more recently, even in our industry, where software vendors have represented that we're secure, everything's all good, redundant data, but then an event happens, you find out that was not the case. Internally, absolutely, we have all those abilities. It's a responsibility we have inside the practice to make sure that we are staying up to date on the latest types or the latest abilities to secure our data. Getting into the AI question around the attack versus the response With data security, there's a lot of different AI tools that we can use to help make our environment more secure and make sure that we're living up to what we need to do. But let's not forget that the threat actor industry, it's a business.

Run it like a business. They are using the same technology to figure out how to break our stuff and how to expose those things. So being able to recognize that fact that securing your internal data environment is not a one time thing. It is a commitment of we are going to keep doing this forever. That's the responsibility on the practice side.

You just hope that your vendors are doing the same thing.

Sanjay Juneja: Yeah. And Clint, to a similar point, like hoping that, you know, vendors are doing the same thing. What are some things as a successful entrepreneur in tech that you make sure to establish that earns the trust and relationships of physicians when it comes to, say there's a specific solution in a platform and it's doing X, Y, Z with this natural language processing, figuring it out and executing, and you're at an 85, 90% internally. When you prop it up, do you have a designated team that just every week is making sure it's performing the same way? You know, is drift an issue?

Like, I know these are all over the place depending on the use case, but some of my concern is some places have the numbers that look great, they get the adoption, and then they're onto the next person and they're not really making sure that the durability or reliability is performing at the same rate. Is that a point of focus, and do you share that with your partners?

Clynt Taylor: Yeah. We definitely share it with the partners. Sometimes it's a requirement. I mean, we work with Humana and Prime Therapeutics and McKesson, and they put us through pretty serious paces. And you just can't fill out a questionnaire sometimes, but then you're going have, well, show me and prove it to me, and so they're going to really challenge you to do that.

So fortunately we have had, and I'm sure lots of companies like ours have had to go through that very, very rigorous due diligence to make sure. Then we maintain a SOC two audit type level of concordance to those things. But now you talk about AI and these are the challenges that every company like us is dealing with right now, which goes back to a little bit of what Jeff was saying, which is now where does that data live? He has some data, customers have data, we get data from outside that has to come in, has to be governed in a way that we know where it came from, how it got in, how often it's getting updated, channels what are in and out of those things and how do we make sure that we're maintaining all that? Because not every group is as advanced as they are.

So what happens when we go into a smaller group who says, we're really having to rely on you to do these things and we have to be able to do that. And look, if there's one thing that probably is the scariest to me, it's what happens if there's a mistake? Right? Because if we all have used tools, it's like, 'I didn't get it quite right, was a PowerPoint I dropped into chat GBT, I can fix that.' But that's not the case when we have a doctor who's been reluctant to use something and now you've got him using it and something happens and they're like, what is this? This is wrong.' We all saw that when Open Evidence had this issue, I think maybe last summer.

Everybody's like, what? You know, was wrong, it missed it and just sent ripples effects. I love the way they handled it. Yeah, well, we fixed it.

Jeff Hunnicutt: Here in the tech industry, the idea of a failure, go,

Clynt Taylor: oh, it's a We'll fix it.

Jeff Hunnicutt: Doctors, if it fails, oh, it's dead to I'm not using this tool ever again. It let me down. I'll never trust it again. I'm not in anymore. So there's definitely more of a, I'd say an emphasis on responsibility of getting it right

Clynt Taylor: after That's our biggest fear.

Doug Flora: I really want to talk about this because what we want to do for this audience, you know, our tagline, we're innovating at the speed of practice, but not all practices work at the same speed. And so here you are, you guys are a juggernaut. You have infrastructure, you've got data team, you've got an engineer at the helm. What would you tell the people out there who come to you and say, listen, I got one EHR, no data team, skeptical board, burnt out docs. Like where does that person start in?

And I'm lucky I'm in a good place that is well ahead. You're in a

Sanjay Juneja: good place. But there's a lot of those. Yeah. Mean, you came from

Doug Flora: one, right? You were a place where it's tough and the records weren't always available to you and you were having to make patient decisions. As your wife, Lauren is sometimes without a path report. And obviously that doesn't happen commonly in 2026 at your place or my place, but I suspect an hour south of my place, that's probably the rule, not the exception. So what do you tell that person who really believes what you're saying?

They've made the decision, I got to get in. How do they start when they don't have a Jeff at the helm? Because it might be somebody like me who's well intentioned and go team, but I don't

Sanjay Juneja: have a data background. And arguably that needs this technology the most. Right.

Jeff Hunnicutt: Strong argument for it. Resisting the urge to reemphasize the tech committee, but it gets foundational, really. Tech committee. That piece is really important, I'm stepping past that. Let's say that that's something that a practice has already embraced and they're well on the road to that.

Identify a single use case inside the practice that you know that you can crush. And you know that AI can help you with that. In the vendor process, we make sure that you're not abandoning the concept of, I want them to be able to do more. But it doesn't mean they have to do all five, six, seven, to 12 things the minute you work with them. Find the one thing that can make a tremendous impact in your practice.

A good illustration here at the community oncology conference, they're emphasizing chart prep and they're showing several vendors that are active and successful in this space that are working with practices. Every single oncology practice has medical assistants doing chart prep. And for most of them it takes, gosh, two hours at the end of every day where they're doing this. The technology already here for you to be able to implement that in your practice. Doc is not going to have to deal with that, but the doc is going to feel the benefits of that improvement.

And so if you can identify a use case like that, that everybody's going to use, you're going to feel that your MAs are going to feel the lift. They're going go, oh man, my day is so much easier now. As you walk through that, if you do it at scale, start to identify the potential efficiencies. Maybe I don't need an MA for every doctor. Maybe the next time somebody resigns and walks away from the practice, can consolidate now it's a pool of MAs who are serving chart prep and maybe lab orders and things like that that they can do with their license.

The potential of the ROI can come, but you need to find something that you're going to feel it in your practice the minute you do it, And that the doctors are going to, I'd say, in an adjacent way, see the impact and go, Woah, this AI stuff's kind of cool. You start to get the buy in of the board with that, but you're not asking the doctor to do anything new. And I'd say that's probably a pretty important thing for

Doug Flora: most doctors. That actually goes back to Emily's stuff too, about establishing trust. You need some early wins. And so we wrapped our last episode and we'll do that with this one too, with a practical take home. This is your homework listener.

And we've got that and Sanjay and are going put that together for every episode. That was what we said is go to a find a whiteboard for thirty minutes. What's our biggest friction point? My, how might we apply technology to help us address that? And then maybe not solve it, but am I skipping steps if I don't include technology in that evaluation for solutions and you guys it's front and center.

So I love that you said that because we don't have a gigantic place, but we have a data team. We've got, I think two or three IT committees that you work through. And I have found that vetting process fantastic, infuriating, effective, and entirely ineffective at the same time because you do have collisions of goals. And my goal might be, I got to find more lung nodules or I got to see more detection on mammography. My IT and my CMIO, they're worried about data security, interoperability, hacking, compliance, you know?

And so, and they're super legit to do that, but it, the clinicians in the room who are listening may not understand why you're not jumping to adopt the technology that the vendor convinced them they needed. Even if it's a legit tool, it might be number 12 on a list of 162 projects you're working on.

Jeff Hunnicutt: And this is an area where I see is a tremendous advantage for the independent physician owned practice So you don't necessarily have the layers of governance and bureaucracy that sit in large healthcare systems. So being able to, being able to work through the process faster, the same dynamics exist. You're going to, IT and compliance. Yeah. And more so compliance than IT, man, they infuriate me sometimes because they'll tell me I can't do the thing that I want to do, but they also keep me on track to where we're not exposing ourselves to risk that's completely unnecessary.

Doug Flora: Put it like that's called a wife. Bless them for putting up with all of us. My wife does that for me all the time, but it's true. They're responsible, credible adults saying, Hey, chill for a minute. Let us catch up.

Jeff Hunnicutt: Exactly. That process in an independent environment, it still happens. The same stressors exist, but you resolve the stress and the conflict, I'd say significantly more quickly because you don't necessarily have to bump decisions up the chain nearly as far to be able to work through that process of everybody aligning and saying, Okay, I understand we started with this goal. We had to change a couple of things because of compliance and IT restrictions, but you got there in three weeks as opposed to two and a half years or something like that.

Doug Flora: I want to ask you, you were on the opposite end, right? And this, I think is really instructive for our team. I've got probably 60 providers. You've got 60 providers in your practice, you and your wife carrying heavy weight. I mean, 28 people or more a day, you didn't have a lot of time to dig in then.

Now you have a completely different perspective. How much different would that have made had you had these tools in your life when you were just drowning?

Sanjay Juneja: It would have made, I mean, just a world of difference.

Doug Flora: You might still be in practice.

Sanjay Juneja: I'm saying it's like, you know, the interoperability was really, you know, a problem and it is still such a big problem in, in, in community centers across the country. I think people don't even realize how bad it can be, especially when you have private groups that you're still sending for GI, pulmonary and all these things. You know, they just don't live in the same EMR. And it's important to highlight that even the same EMR does not mean you have interoperability, because that has its own challenges. But what I've learned is the fix isn't as quick as you would think.

And I think we were just alluding to that. I would love if you all could actually maybe color in or detail exactly what that cost is. And I don't mean paying for the actual, you know, say product or solution or platform, but really in the time and effort and load, like you have almost an FTE that may have to have the entire project like under their belt for sometimes six months, as I understand it.

Doug Flora: Or 10 of them.

Sanjay Juneja: Yeah. I mean, exactly. Just, just to start the process of actually getting to the solution. Number one, why does it take so long? And number two, do you see it getting any faster?

And if it is, how long is it going to take? And, and I guess third, my concern is, is it contributing to people having some resistance on adopting big solutions like you're saying that my wife and I needed at our practice? Because the lift is so high that you don't ever want to have to backtrack and therefore you don't make a decision or commitment because that has its own, you know, analysis.

Jeff Hunnicutt: It's funny you're describing a situation that we're actually dealing with right now in our practice at Highlands. Think it's a critical mass type of growth thing. Most practices, if you rewind back a few years with Highlands or gosh, go back to the practices that I've worked for in the past, when you got 10 docs and two fifty staff members, something like that, everybody wears a ton of different hats. So there's no single person inside the organization that is the glue that is pushing these projects forward. There's no project director sitting in that executive or not executive layer, but management layer.

It just doesn't exist because you can't this is a luxury you can't afford. So everybody just rolls up their sleeves and says, You're going take that, and you're going to take that. And hopefully you've got somebody who's good at email and CC ing the right people and keeping things moving, but that's kind of how practices solve for it. At Highlands, we're at that inflection point where the projects we're doing have enough widespread impact into the practice or even sometimes outside of the practice, building new buildings and starting new service lines. We're feeling that stress point inside of our group now.

Got, God, we've to have somebody whose job is to be the glue. Have to be the person that keeps everybody on point and moving in the direction because it's just harder to do that when you're at that scale without someone that's assigned to that responsibility.

Clynt Taylor: So you have led in the implementation of lots of different types of technology projects over the years. I'm curious, you also have adopted a couple of new AI projects. Is there a difference? Was it as hard to implement ChartPrep as it was some other project before? Is it changing?

Is it an easier? What are you experiencing?

Jeff Hunnicutt: It depends on the nature of the project. If we use ChartPrep as an example, most of these products that are in ChartPrep are functioning as a Chrome extension. So the end user experience is that they just got an enhancement to their EMR. They may not even recognize, there's no, not even necessarily overly obvious branding in the extension to say this is this new tool. Maybe more the icon might be the only distinguishing factor to it.

The speed at which you can implement it, if all you're doing is you're altering your group policy inside of your server environment to actually add this Chrome extension to certain user accounts, the implementation, the installation is very seamless. The user experience is seamless as well. So your speed, your speed to getting things back up and running is immensely quicker than say like, yeah, bringing on a new EMR or some new software suite. Have to bring in people for training. You have to worry about HL7 interfaces and building an infrastructure around the product.

Several of these newer tools are just more

Doug Flora: Can I ask, so you know how I spend my life now? I'm sort of proselytizing and I hope people aren't tired of hearing how urgent this is, but I speak from the patient experience. The dread that I felt when I had cancer nine years ago, and I'm cured, we'll probably mention this every episode because I want to bring it back to the patients that we're serving. The burning platform here is the urgency, right? We don't have a choice if we're going to do this or not.

Now we have to figure out how to do it safely and responsibly. And I guess my frustration when I think about my perspective as a patient is I know we can do better. We talk about the moral injury. We talk about the gap between what we know we can deliver and what we're delivering right now. And I know during COVID how it worked.

We're going talk a little bit about urgency. During the emergency, we got stuff done. Clinical trials open in two weeks. You've got things through. Emily last time talked about she got on Amazon, she ordered a tent.

That's where they drew blood. That emergency went away, but patients are still dying at a rate that is unacceptable to me as a doctor, as a patient, because they're not having their nodules detected by an AI enabled software patent. So I get very frustrated and probably speak out of turn about my sense of urgency is different than other sense of urgency because that protocol, that package that you've purchased that sits on the legal's desk for nine months, we probably lost fifteen patients. They died needlessly because of our committee structures. And so can you help us as a group, community oncology in general, because you are more nimble.

How do we get that sense of urgency back into the boardrooms, back into where the decisions are made around the doctors to say, The delays here are not acceptable and they're actually consciously causing harm with this sort of, we'll do that next quarter. Because the tools are now, many of them are published, randomized phase three trials saying if you do remote patient monitoring, you extend survival. And we would do that in a heartbeat if it was a new drug, We clap at ASCO for that and we change our standards of care. The same tool exists for early referral for a metastatic lung cancer patient to palliative care and we're not automating that and it's not happening. How do we build that sense of urgency and responsibility because it's a hard balance, right?

Jeff Hunnicutt: So convincing a team of people that are actually caring for patients, the issue that we need to move forward on some new tech is different than convincing a board. Thankfully, the groups that we're working with in independent oncology, it's a board of physicians who are in there and they're practicing and they're seeing these patients. It's easier to be able to come in and, and I don't know, the pitch, to speak, on something that's going to benefit clinical care, because they're the ones that are going to be affected by it. The patient that they just saw, they're thinking of the face of the people that they just saw. That's going to make, make them better or make their patient experience better, I should say.

So I'll use ePro as an example. Hyland's was an early adopter of ePro all the way back in the summer of twenty twenty. EPro itself is not a new concept. I was working on what you would call ePro projects back in 2006 with tablets in the waiting room of taking eCoch status data before they got to the exam room. Same concept, but just obviously a smaller gap between when the patient is giving the feedback and when they're actually seeing the doctor.

Well, now you put that on a cell phone and it could two weeks before their next appointment, they can let you know. With ePro, it was fairly easy to get the board of physicians to be able to get on board with that because they could see the triage experience now is just, it's far from optimal. And they know that by the traditional call in, leave a voicemail, you're good at it, maybe you're getting back to them in four hours. So coming to them and saying, Hey, I see a solution here. The data shows that on a macro level, that we're going to make a huge impact, but the the immediate care, as far as how quickly they can get ahold of somebody, how quickly we'll get back to them, how quickly we'll resolve their issue.

All that is, is trackable. You can see the whole thing. You can see exactly how people are doing. You can see who your rockstars are inside the triage department and who are the ones that are that need a little nudge to be able to care for those patients better. So they'll feel it because their patients are coming in and saying, oh man, I love this new app that you guys gave me.

I was able to get ahold of insert name of triage nurse here and she helped me right away. Man, it's great. The doc instantaneously is going, okay, that's a good thing. Now, if you can layer on billable codes and things like that after the fact, well, now you talk about ROI and that's an important part of the business decision, but the clinical buy in is already there. So as that product implementation evolved, the billing codes came later, really.

It was your RTM, your CCM, those came CCM came a year after we started, RTM came three years after we started. But now the ROI is right side up to where you're actually making revenue for the practice, which you kind of got to do, but you build something that the docs were bought into.

Sanjay Juneja: And was that on a leap of faith that if it's going to help patients, eventually the billing will make sense? Or were you aware, like were you investing in it knowing that was down the line?

Jeff Hunnicutt: We knew, so the timeline with that was lined up so that we knew OCM was ending at the end of twenty twenty one. It had the one year extension during COVID. You couldn't bill chronic care management for Medicare patients during OCM. Was a double dip. We knew that there was an endpoint that was not too far in the future where we would be able to bill chronic care management and see their ROI come home.

The clinical differential for the patient was something that convinced them, Let's pull the trigger on this a year and a half before it'll actually cover itself expense wise, believing that that part will come, but that you'll benefit in the interim.

Sanjay Juneja: Yeah. To the point of patience, this is kind of a, maybe a curveball of a question, but an important one. And that's, I get often asked patients and their feelings about AI tools being used by their physicians. So when y'all are making your decisions, are you somehow sampling patient sentiment about a potential tool like chart prep, because I know perhaps a patient isn't as well read on AI, or maybe they're better read, who knows, about the quality of the chart prep, etcetera. How much consideration should go into the institutional decisions for just a patient's comfort level with the use, not necessarily of what is downstream directly in care, but the utility of it.

How do you navigate that?

Jeff Hunnicutt: So, technically, when you talk about a declaration to the patient that AI is being used in your care, we've had that declaration in our new patient paperwork for seven years. Going back, AI is an umbrella term that covers stuff that goes back to the 80s technically. I mean, machine learning and pattern recognition technically is AI. Now, of course, we weren't in the first go round of EMRs for, gosh, I don't know, the mid-2000s. No, you didn't think about putting that into that declaration inside your new patient onboarding process.

But once you get around to, gosh, yeah, 2019, 2020, some of the tech evolves to where patients, the concern around AI was already was already happened back then. So addressing it head on, putting that into that new patient process so the patient could see when they're reading their paperwork, AI may be used for certain applications during your care. You've already been letting them know that that is potentially part of the process. I'd say it's very, very little concern that we've had over the years, if at all, where people have said, I don't like that. I don't want to be a part of that.

As the tech evolves and you're using it in more use cases, maybe that changes.

Doug Flora: We started earlier talking about AI is not necessarily just a product. It's a capability that your practices earn. I think the ePro solutions that you guys have, remote patient monitoring are a great example of applied science that you can see a discrete, this helps my patient. I think as a community, are talking a lot at this meeting. We're a co op for you guys who didn't catch that earlier, moving more complicated care into the community.

So bispecifics have arrived in earnest, right? We have a lot of our centers, a lot of our partner centers are giving outpatient bispecifics. These patients can become critically ill, predictably critically ill. And I think not using these tools at this point is probably going to be looked back at as very barbaric, very malpractice equivalent because you're, you have the tools they're there. You can catch the patient the first hour of their fever with a wearable.

You can act on it twenty five seconds later when an agent responds to the fever or a nurse clinician or a nurse practitioner who's working remote 20 fourseven, here's patient spike temperature. You get that patient on the antibiotic in hours instead of four hours before the phone call comes. I love some of these. And again, we're not talking specific vendor products or whatever here. It's a larger discussion.

May the best product win, but to me, that's a very compelling use case. And I don't know why we've been dragging our feet on that. I think it's super easy to do business intelligence, right? Prior auth, everyone's like, yeah, I could do that. I hate prior auth.

This is probably as simply introduced, but directly affects the people when you have a physician board, that's where community oncology wins. I don't have a physician board in my practice. We have administrators and people that run companies and they care about our community. They care a ton about our patient, but they may not feel that urgency that you guys feel because you're in the room.

Jeff Hunnicutt: Some of that technology you're describing in the bispecific space that exists today. There are practices that are live, that are using ePro, but have a tailored approach to the bispecific patient. It's using these innovative technologies to flag those and surface them back to your after hours care team to make sure that it's differentiated. Hey, this is a bispecific patient. And here are those unique identifiers or unique elements, I should say, that you're going want to pay attention to.

All that is the text taking care of So that, so that you're out your after hours nurse or wherever it is, the doc knows, okay, this patient's different. I got to treat them differently. It's already there. People just need to be, I guess, more embracing of the idea of using it.

Doug Flora: And I talked about not being vendor y, right? For you guys, it's going to be so frustrating because you have a tool that you know can deliver these outcomes. Again, randomized peer control, peer review trials in my journal, I can show you the survival advantage for these tools and we're slow to adopt and it must drive you guys nuts because you can't get in the room with the decision maker. If you can, it's stuck in legal for a night. Talk about the way you approach this.

When you have either a skeptical doctor or an overwhelmed doctor, a doctor says I have no budget or a doctor says my IT team says no.

Clynt Taylor: Yeah, those are a lot of different types of challenges. I think the first one that you mentioned has a lot to do with just understanding that you rarely have an opportunity today, never have the opportunity today to just come in and replace whatever's all there. So you have to be you have to think about the context of what's already there. What EMR do you have? What tools are using?

We had a great conversation. You're talking about a wearable. I had a great conversation yesterday with a wearable company who came in and said: We're super excited about what we're doing. We have practices adopting this. Tell me about it.

Tell me about the workflow. What happens? Well, then they get a message that says: Here's something you need to act on. Then what happens? Well, an MA needs to take that and there's a protocol that follows this.

There's a test to be ordered, there's a follow-up to call, blah blah blah. Oh wow, the excitement then becomes what if we automated that whole thing? It's how do we find ways to support the use and adoption of those tools that you already have, that you've already invested in. I think that's probably the biggest challenge because there's a lot of stuff that's already there. It's already in the practice.

Sanjay Juneja: I have to ask, you know, thinking about this automation and really what the world could look like in a year or two, how do you approach dependability and the concerns of really, depending on a tool or automation so much so that you could have just an enterprise shift where a company gets acquired and decisions get changed. How do we account for that? Should there be policy that de incentivizes having these tectonic shifts that can all of a sudden just make an entire ambient company liability for the physicians that finally got used to it, or the chart prep? In your MA example, you did consolidate your MAs, everything like that happened. Then all of a sudden, one acquisition or a company failure or something, they're out, and now you're desperate to hire MAs.

You could see how that's a mess. What is the preparedness or backup plans for these things look like?

Jeff Hunnicutt: I'll go on that one. One is something that I think practices have to be thinking about. You're going back to what I was saying about that vendor interviewing process, part of it is you to give some time behind closed doors to think who's likely to buy this because most of these products are not going go public. Most of these products are things that have a five to seven year ramp life where they're looking to have some sort of exit and somebody's going to buy them. I'm going use a common one, AmbientScribe AI technology that sits in an exam room.

It's a beautiful compliment to any EMR. The likely pathway with these is that they'd be acquired by an EMR. It doesn't mean that you don't move forward with the tech, understand how things might be influenced if an EMR acquires that. Think about some of these other things. Could they be acquired by pharma?

Could they be acquired by a large private equity institution? Could they be acquired by one of the big MSOs? What does that look like when it happens? It doesn't mean don't do it, but you should understand the potential ramifications of a deal before you actually dump all your eggs into that basket. If nothing else, you're prepared for what life after the deal might look like.

Sanjay Juneja: Yeah. I just think that's so important and something that maybe isn't considered enough because we're talking about change we're talking about adoption, we're talking about all these things. And then the rug gets pulled under you.

Doug Flora: Happened this week in this lobby at the Swan, where are we? Dolphin Hotel at Koa. I watched in the lobby all week long as these vendors were starting to click together like Lego blocks. We keep saying that there was a major acquisition announced Friday. I probably won't say the name because this is Evergreen, but it was two companies that really compliment each other.

Now they control probably the top five centers in the country with entirely complimentary skill sets. I love that because that's what we talk about. Point solutions are really hard. We can't buy 200 programs for 200 problems and we have 200 problems. So I think we're going to see some significant acceleration of consolidation of these companies.

And I know because on our side, a lot of them are approaching us as early adopters thinking that we're agile or easier prey or however you want to put it, depending upon your perspectives. And I keep telling the same thing like, yeah, I think ambient scribes are fantastic. My doctors use them. I really wish it was attached to four other things. And I think those companies know that, they've heard that from you and I because they ask.

And I know that there's some of that consolidation going on and clicking together. Maybe as we look forward a year or two from now, what is Highlands working on that we will be reading about that we're going to be jealous of? Usually I'm pretty jealous.

Jeff Hunnicutt: Well, mean, the thing that we, I'd say is probably the newest and most exciting

Doug Flora: Besides your building.

Jeff Hunnicutt: Well, in technology, the thing that's the newest that we're actually live on now that's been generating a lot of buzz, actually even here at the conference, has been a Genetic AI inside of the switchboard or inside of the operator arena. You know, AI taking calls, necessarily a new thing, but not necessarily something we've all been excited about. I mean, I'll raise my hand first. How many of us have been on the phone call or a phone call with an airline and you're talking to AI and you're screaming customer service, you know, into this thing, to get past the AI? What we're seeing is that that tech has evolved in leaps and bounds.

We turned it on, it's very new. We just turned it on in February and our use case was to turn it on for those midday overflow calls. When you got too many people calling in, your queue can't handle it, most practices their phone systems will dump to, it's probably the same service you use for your after hours service, They take those calls, put in a ticket or put in a note and send it back to you. It's perfect use case to try that AI first twenty to 30 to on a busy day, maybe two hundred calls on a Monday. That's a bad one.

But what we've seen is that when from the mini you turn it on, I'll give it to an example because it's so cool. I love it. There's this guy he calls in and the AI picks up, Hi, I'm Kathy. I'm your virtual assistant with Highlands Oncology. The voice sounds great.

Sounds better than Siri ever, ever can. And it's like Predict your voice

Sanjay Juneja: or you

Jeff Hunnicutt: can. Predictably the patient, it kind of pauses for maybe a second and a half or so and says, and responds the way you and I would probably respond to an AI engine, right? He says something like reschedule appointment, not conversational. And the AI answers and says, yeah, I can totally help you with that. I'd be happy to help you reschedule your appointment.

Can you give me your name and your date of birth? But says it as fluid as I just did, The guy goes, Oh man, I thought you were a robot. He starts talking to it just like you and I would talk and it does a flawless job. In our workflow, our operators, what they're doing is they're taking the call, they're entering a ticket and they're assigning it off to a place where it goes. So voicemail really doesn't exist in our practice.

Allows us to track everything. It's great. Well, it's, I mean, in the IT side of me, going back to, gosh, the early 2000s when I was doing IT for oncology, I can't tell you how many phone calls IT gets that says, essentially, it's the dog ate my homework. I didn't get the voicemail. We have no way to actually prove that that's true or not.

Well, if there's no voicemail and it's all ticketed, then they can track everything. It's great. But back to the AI example, the AI is taking all that information from the patient. It's filling out the ticket and assigning it where it needs to go. On the first day, I think it was 85% success that we saw on the first day.

Within a month, it was up to ninety four percent. I think it's even higher than that now. What does that do as far as the possibilities in your practice? Your after hours call? Yeah, everybody enlists an in person after hours call center.

Does that need to be around anymore? The answer's probably no. And as far as scale into the future, my practice grows at a rapid rate. And without that kind of tech, I'm going be hiring operators every year over and over and over again. So can I slow that growth?

Is there through attrition? If somebody leaves the practice, do I need to replace them? These are all be they all become possibilities because of implementing innovative tech.

Doug Flora: It's critical. We talked about the burning platform a couple of times and we don't have enough doctors. We don't have enough nurses. We don't have enough MAs. We have too many patients.

And it doesn't take a genius to figure out what we have to do. We have to scale everybody. That nurse needs an extender. That MA needs help. The doctor needs a copilot.

I think these tools are starting to emerge as viable, realistic opportunities where we're starting to understand the early wins like you've just described. For me, was capacity management. We, we really moved into that. We've been pushing for it for a long time, but we're an Epic shop. So I got the Epic version, which wasn't my favorite tool.

You know, to our great surprise, after a really painful month, it has fixed the store. And so now we don't have that peak. It's flattened between ten and two. The nurses are getting lunch. The teams are happy.

Pharmacy is complimentary. And again, it wasn't like it was a major, major thing, but one tweak just eased the pain of maybe 50 or 60 patients a day who were getting stuck at that peak. We probably treat, I think we have 84 chairs running at any one time. We're treating 500 in my building, but maybe 50 of them were truly inconvenienced because they had to wait eighty minutes for a drug because the pharmacist got 10 at the same time. That doesn't happen anymore.

So, those are the wins that I want to communicate to our audience so they understand these tools, when appropriately vetted, appropriately teamed up as a synergistic partner, these are the problems I'm trying to solve. Can you help? We narrow that gap that bugs us a lot.

Clynt Taylor: A lot.

Sanjay Juneja: Yeah. And I want to underscore something that you might've missed. And it was that Jeff said, we started at 85% and got to 93%. And I think this concept of this iterative, reiterative process really needs to be understood by physicians and administrators. It is a not out the box, this is how it performs, this is whatever.

It is iteration and reiteration. I don't know if you can characterize what that process looked like. Was it the vendor that was like, okay, looking back and saying, okay, we can make these agents smarter or more narrowed. Was it you all? Was it collaborative?

How did you exactly do that?

Jeff Hunnicutt: Collaborative. And honestly, it had to be. There's an excitement in our practice, I think that does breed from that culture of loving doing stuff that's new. I'd die in the vine if we couldn't do new stuff, but that's trickled down. When we went live, I was curious as all heck.

I was getting some of those voice samples sent to me so I could hear what it sounded like myself. But before I even got them, the manager that's over the top of that particular, like the administrative services operator team, she was reviewing every single call. With interest. Mean, see, okay, is an accurate skepticism, but also extreme excitement. I mean, she doing backflips, seeing how cool that tech was.

So you're energized by some of these calls, like I told you, there are other ones that are even more impressive on day one. Recognized the patient needed a peer to peer and it got a scandin and it handled it beautifully. I thought it was going to puke on its shoes when it got that one. Still But maintain your enthusiasm, or I should say, temporary your enthusiasm by looking at how did it really perform. Looking at it and say, man, okay, 85% on the first day.

Wow, I did not expect that. But there's no way in heck I'm going to turn that thing loose on my entire volume of thousands of phone calls in a day on 85. No, I need ninety, ninety four before the next jump. I want 99 before I let it into the phone queue to actually be answering calls like as if it was an operator. But that process, have to be able track the data.

You have to be able to review the successes and failures, analyze the failures. Where did this thing go wrong? You know, the gap in its response is too long here. It interrupted the patient when it was speaking here. So adjusting those little tweaks, we have to do it in tandem with the vendor.

Sanjay Juneja: And you went back to the vendor and told them that?

Jeff Hunnicutt: Oh, we were pretty much listening to the calls together for the most part.

Doug Flora: Everyone in the audience is going to be texting Jeff to find out what, wait, who, what, what did you buy? And we won't do that here guys, we are brand agnostic. We are, we're not spokesmodels here.

Sanjay Juneja: This is the point is like, there's this malleability to this AI technology that I really hope is received by anyone listening to this. Like, and there's a dopaminergic reward system that comes with participation in that optimization of that thing. Yeah. It's not, it's just not the way we think of technology ten years ago.

Clynt Taylor: And it wasn't a failure at 85%. Wasn't like was

Doug Flora: far better than what we started with.

Jeff Hunnicutt: I was excited about 85%, but I also understand the realistic nature of that scenario is that I can't be satisfied with 85%.

Doug Flora: We're almost out of time. Of course, could go another hour and we probably will after we turn off But the is why we want to do this series and that's why we're going to continue to build on this. Lucia Gordon's next. We're going to talk about even bigger, more innovative things at the larger practices, but I think the difference here is the mindset shift that hopefully the audience is starting to sit with and understand that you might have arrived at this thinking about AI as something to buy. And we talked about that.

That's not the point of this exercise at all. I want you to leave with the idea that it's something to build towards and you have to earn it. You have to do a little grinding. It's going to be some iteration and some frustration, but my example with our capacity management for moving infusion chairs around with basically playing Tetris and filling those holes. Your example with managing that, the pre charting, the RPMs, the EPROs, I hope the audience is starting to understand that these are some of the solutions, but choose your own problems.

I want to leave with what Sanjay and I are going to do each episode is give you guys homework. We want to make this a practical tool and I write this down so I don't forget. Listener call to action. If you want to be like Jeff. I don't know if I've ever said that before.

I know we all do. We all do. I've my friend for a decade. Please. All right.

Listeners call for action. Pick the three most important clinical or operational questions your practice should be able to answer with its own data and find out if you can actually answer those today. If you can, you are closer to AI readiness than you think. And if you can't, you just found your first project, right? Just start there.

You don't have to buy anything. You don't have to do anything. You don't have to upscale, upsell any of the stuff, but either way you're moving in the right direction. And if you can answer that question, you are close, You are close. And we're going to start building out Sanjay and I, we have a very good platform for we're going to try and help people understand AI readiness and sort of a pyramid of AI enablement from, I know how to use Copilot, which is not my favorite obviously, but that's my enterprise solution to, have a fully agentic enterprise.

And we'll be getting to that probably episode six, episode seven. So thank you all for joining us, Jeff. I knew it would be a kick. Awesome.

Jeff Hunnicutt: Have

Doug Flora: embraced my inner nerd again, and no one's surprised because it doesn't live deep, does it? It comes out quickly. Join us for the next episode. We're very glad to have you and we will continue to practice at the speed of innovation.

‍