This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

 Today on This Week Health.

(Intro)  so it shouldn't be you saying, Okay, chat GPT is out. Hey, we think this could really change something for, automation. Can I get some funding to go try? It's the other way around. You've got to be constantly reinventing and trying to displace some of the friction in your business .

  Thanks for joining us on this keynote episode, a this week health conference show. My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of this week Health, A set of channels dedicated to keeping health IT staff current and engaged. For five years, we've been making podcasts that amplify great thinking to propel healthcare forward. Special thanks to our keynote show. CDW, Rubrik, Sectra and Trellix for choosing to invest in our mission to develop the next generation of health leaders. Now onto our show.

(Main)  do want to thank everybody for joining us today. We're going to be talking about practical AI. We did an AI discussion earlier this year with several academic medical centers, and they had funds coming in from the federal government and all sorts of other stuff. We're going to be talking AI on a budget.

We're going to be talking AI in very specific practices and how we are stepping into this world. And I will introduce our panelists . It's David Baker, CIO Pacific Dental. Reid Stephan, CIO, St.

Luke's out of Boise, Idaho, and Lee Milligan formerly with a bunch of companies. As I like to tell CIOs. At some point, you will have my same title, which is Former CIO. And the title you currently hold. want to encourage people, if you have questions, go ahead and pop them in the chat, but we have received

So many questions, we won't get to all of them by the time this this webinar is over, but we are going to try to get to as many as we can. But I'm going to start the discussion with a question I've been throwing out to CIOs for a while. Scale of 1 to 10, what kind of impact do you think AI is going to have on your specific field?

And Reid, we'll start with you. What kind of impact will AI have on healthcare?

Yeah, I think it breaks the scale. We've talked about this before. It's an 11. And I think that's, I'm always a glass, always full kind of guy anyway, but I just think the speed with which this has come into kind of our radar and then just the things we've learned over the last year, I think this is going to be incredibly impactful in a good way.

And so I'm very bullish and I'm over a 10. It's an 11.

Lee, we're going to keep you focused in on imaging. AI, is it going to impact imaging at all?

No, not much. No, just kidding. Yeah, it's a complete game changer in this space. There's, 450 plus companies right now that are working on AI solutions specific to medical imaging.

And they're all taking a little bit different slice of the pie. And ultimately, five years from now, the landscape will look very different than what it looks like today. I actually agree with the number that Reid put forth.

Love it. And David Baker talk to us about AI. A lot of us aren't, on this call, may not be as familiar with dental.

Although, I'll give you a couple seconds to tell this group of listeners why it's important and why you went to Epic. So go ahead and tell us why it's important and tell us how AI is going to impact the dental space.

So for us it's pretty huge in a number of areas. I think we've been at AI for a while, right?

So I think it's definitely a 10, but we've been working in this space for probably three or four years. It's just suddenly become super hot because some of the more, generative AI tests and obviously chat GPT just accelerate a bunch of conversation around it. I think the tools we've got now are amazing if we can liberate them right with the wealth of data we got.

So for us in the dental space, we started with AI in imaging funny enough. So we've been heavily down that road for a few years now. We're just bringing that to some really exciting techniques around some of the preventative diagnosis there. So it's yeah, it's 10 for me, I would say for

sure.

All right. So give me the quick elevator pitch on why. Pacific Dental went to Epic.

Because we were on a budget and we we wanted a cheap system to implement. That's where you wanted me to go, wasn't it? We why did we implemented it? Cause we're now not just a dental services organization.

We're in a medical business as well. So we're bringing. Primary care physicians together with our dentists in the same office, ultimately, and looking at what we call the mouth body connection and, complete body care, essentially. So EPIC was a big part of that for us because we were finally able to bring together what have historically been Truly disparate data sets between these two industries.

Not only that, it was just, a superly mature system in the space in the back end and administrative side of the house. And we felt we could leverage some amazing operational opportunities in in moving faster. Things like, real time eligibility and and just the automation of the many pieces of RevCycle space as well.

Cool. We're going to spend over half of this webinar on your questions that have been asked. But I'm going to keep going on just setting up the conversation a little bit. Lee, I'd love to hear you talk about current approach that organizations are taking with regard to medical imaging to AI.

What are some of the specific areas they're using it? How are we getting the organization ready to make these kinds of changes or adopt this kind of technology? and where is it going to be used in the near future?

I think first we should talk a little bit about the impact to the physicians who traditionally have delivered this care, right?

This can be a very scary and daunting idea, right? That some, computer is going to take over their world, take over their job, take over their meaning in this space. And I think it's actually a legitimate concern, right? Because it does some things significantly better at this point.

And so that's actually part of the conversation is really talking with physicians about what does that new framework look like moving forward. And so I think from a health systems perspective or an imaging perspective, you really have to ask very practical questions around what it is you're getting when you attempt to implement AI. if you're a typical health system or imaging company, you can't be spending a lot of your time developing it on the front end without getting the operational impact right now, right? Your board's not going to stand for that. So you have to ask good questions about is there an operational efficiency that we are gaining by implementing this?

Are we improving something clinically by implementing this or are we mitigating risk? By implementing this. Those are really the three things that, that many companies are looking at to identify whether it makes sense to put something in place. One thing I will mention from an imaging specific perspective.

And by the way, obviously I'm quite biased on this topic given my background, but imaging is so amenable to AI, right? Cause you get millions and millions of. Images, which are really just patterns, right? And then you have millions and millions of discrete fields, ICD 10 codes, right?

That you can match those up to. So the ability to apply AI in this field is really tremendous. But, if you think about it, in order to implement an AI solution from the FDA perspective, you actually have to do it on a diagnosis by diagnosis basis. So it's actually a very painful process to go through for a company attempting to do this, right?

So let's take chest x ray, for example. They can't just say, I have an imaging solution for all of chest x ray. It's chest x ray specifically looking for pneumonia, or pleural fusion, or collapsed lung, whatever that might be. So it's a process to go through that. Many of these companies, what they do is they really join with a a vendor, and they co develop to some extent, so they'll start with something that's already FDA approved, and then together, they will look at additional indications.

And they'll work on that together and then submit that together. So we're in the middle of that process right now, and it's really exciting.

Are we going to see significant changes in the way that images are read, in the way that images are processed, in the way that physicians interact with diagnosis from imaging?

will it look fundamentally different in five

years? I can tell you a couple of quick examples. So let's look at long bone x ray. that's something we implemented a year ago or a little bit over a year ago. And initially the sensitivity and specificity was under where we needed it to be.

We did some tweaking on that and eventually got to a spot where it is as good or better. Then most of the folks who are reading the films. And so that changes that dynamic tremendously, right? The way I viewed it is, in the future, physicians will have a slightly different role in this space.

Because instead of having an individual look at every single thing, it's my opinion that AI will look at everything, and then a fraction of that, a portion of that will be bubbled up to the human being, the board certified radiologist in this case, to review something that kind of falls out a bit, whether it's an abnormality or something that doesn't quite fit into the normal algorithm.

But I think given the volume of work out there to be had, And the finite number of clinicians that are coupled with, right now, this amazing surge in AI, I think the model is more like a checkout lane. I think I mentioned this before at one point. in the old days, you'd wait in line at Walmart, and there'd be like 25 lines to get out, and there's a human being at each of those lines checking you out.

Today, they have a few of those, but for the most part, you walk into a space and it's self checkout, and you have one person circulating in case there's a problem, right? I see a similar analogy within the imaging space as well. Everything looked at by AI, and then a fraction of that looked at by a human being.

David

Baker, we're going to have you go next. that in There's a receptiveness to this technology. It's been around for a little while. There's a receptiveness to it. You're not. Pushing it on them, there's almost a pull that may not be as evident.

You and I worked together at St. Joe's. When working with doctors, a lot of times we had to push a little bit the technology and find :champions and do the whole organizational change. it the same in the dental practice, or do you find people are pulling it through?

It could be a little generational, right?

Our company's very young in terms of the average demographic and owner doc. So I think they're very receptive to the tech. We've always been pretty tech forward, we'd say proven technologies here. From, the milling equipment to 3D CBCT scans now being the standard of care for us, which is unusual, I think, in the wider dental community.

In the most sense of Paul, in the AI, there's been, folks have been on top of it. They've been pushing me hard to get the product out there, which we started to, there's a lot of folks in the space now. Medical, we've been doing it for a long time. To Lee's point, the FDA approval was the sticking point for us.

We ended up doing a joint venture where we stopped developing some of our I, some of the IP around the AI product that we had for our innovation lab and sold a piece of that and then went In with the manufacturer who makes out large CBCT units, which are super costly and help them catch up with some of the software and became like, their first largest customer deploying it.

So there's big ball because if you think about it, when you're in the chair, right? So the docs want it when they realize, okay, it's not going to take my job. But my job is to be in front of the patient and give them an excellent level of care, right? As they're ramping in their career, the stuff that it's pointing out on an x ray, for example, is super helpful, right?

It's quicker to get to the, okay, where do we see some issues with the scans that we've got? And then patient side, you think about it... It's very traditional business dentistry where you have your local practice, the local person that you know that you recommend to your friends, and it seems like a more personal kind of interaction, right?

So when we're building these practices, and obviously the patients that go there, there may not always be this immediate trust. So when you're told, hey, you're going to need a root canal, or you have a cracked tooth, it's okay, great, I'm going to get a second opinion, because I was only told I needed a filling by the last dentist.

So there's this kind of bounce around. So now when you combine the You Actual information in front of you is an amazing 3D, technicolor looking scan of your skull. And you're able to zoom in on our touchscreens and talk through what you're seeing. And oh, by the way AI backs this up within a 90th percentile of accuracy, what I'm saying.

I think it becomes really powerful, both patient side and doc side.

Yeah, it'll be interesting to, see how this gets down to the patient and how we experience it. Reid, I want to talk Lee set this up really well with the various things, are we looking for equality?

Are we looking for efficiency? how are you preparing your organization? In what areas is your organization saying, hey, It would be great if AI could do these things.

Yeah, so maybe if I can just like wax philosophical for a minute to lead into that. And I think David made a point earlier that's well taken, which is like AI as a category is not new.

What's new is, in the last year, the ushering in of large language models, this application of AI, really a killer app kind of nature of it, that's made it much more accessible to a much broader segment of population, almost on a daily basis. So what we've learned as we've gone is that.

I talked about being an 11. I've had to understand my excitement. I have to meet people where they are. And so there's some understandable and natural fear and trepidation. So we've had to be intentional about how we think about this and talk about it. This is not about imitating humans.

This is really about augmenting and elevating their productivity, their creative capacity. It's not about jobs ending. It's about reimagining what those jobs are, which is going to require some adapting and some reskilling for sure. And this is, it's also not a race. I think, I really love this conversation because it's creating this collective journey experience.

And I think that in this Error we're in, collaboration is going to trump competition. But this is not back to why I'm at 11, this is not just a trend. This really is, I feel like, the dawn of something new. And the paradox is that it's like there's endless potential for how we might imagine and think about the application of AI. it's also total uncharted territory. And so I think there's... appropriate need to proceed with some caution and making sure that we're being responsible. So for us, we're looking at those areas where it can do just that. It can take work that humans are doing today, that it adds value, but it doesn't fill their joy bucket.

And so it's work that we can provide some, a digital assistant or some augmentation that It still takes computational human gray matter to do, but if we can offlift that from them, then they can use that given back capacity to work and provide even more value, but also be engaged in work that it's more meaningful and it's more enjoyable for them.

There really is a possibility to find, I think, that sweet spot. And I think as we go on, we can talk about specifics. But that's at a high level how we think about this philosophically and how we're trying to approach this journey.

All right. job is done. I'm to start going through the questions that came from the group.

And you'll see why, because they're very direct questions. And I think it's questions you're probably getting from the people that you interact with, who are looking to you for leadership. Number one use case for generative AI that moves the needle for workforce challenges. Reid, I'm going to start with you on that one.

Yeah,

so I'll share a practical one. And again, this isn't you're not going to walk away thinking, Oh, I've never heard of that, but I'll share our journey. So one of our approaches has been, rather than go out and trying to out of the gate, create our own private large language model environment, we're dabbling and learning, but that's not our primary use case.

Let's align with these vendors, these partners that we already invest a lot of resources with and understand where they're at in their AI journey, and then how do we take advantage of that. So let's start with Epic. InBasket, as we know, is the bane of a lot of folks existence. And with Epic, The ability to weave in AI and to be able to auto generate responses to in basket messages.

The working theory is that's going to be a huge provider satisfier. So we're a few months into our pilot slash proof of concept, and it's been really interesting. There's been good. I think overall the providers can see the value. What we've learned is the initial tuning was provide a general response.

Don't give any medical advice. And so what that has looked like for us is, it's good at answering really benign type questions, but anything at all that strays into the medical arena, the message will end with call and make an appointment to understand this better. So it's just shifting that burden like down the line to somebody else.

So now we're tweaking that a bit and saying, here's the guardrails in which you can offer some medical interpretation or advice. Because again, That's not going to automatically go to the patient. You're going to have a provider who reviews that and looks at it. So we're going slow to go fast, but we're finding that there is definitely value there.

And again, it's going to augment and reduce some of the burden that exists today. With communication, it's important, but we can use AI to, to be more productive and to be more creative. I'll just share one other thing. One of our docs said, The one thing that it's really good at, it is consistently more empathetic in its tone than I am consistently, because I'm a human, and like the ebb and flow of life and my fatigue and attention, I just, can't consistently be as empathetic as the tool is, which is...

Really fascinating.

So when a doctor rolls his eyes, it comes through the keyboard. Yes. Is that true, Lee, that happens from time to time?

Not to Lee, but

other doctors. Yes.

I'm not going to have all of us answer every question, otherwise we're not going to get through them, but either of you want to comment on generative AI?

I have one thought on that. There's tons of use cases and I'm glad Reed started with partnering with a vendor because I think That makes the most sense given constrained resources. But the other one that really is striking to me is the idea of creating summaries, right? Shift change summaries are huge, right?

There was this huge effort, for the last 15 years to try to pare down all this stuff. That's in the note or that's in the, just sitting down at the dashboard of Epic or really any EHR, right? It's like sitting in the cockpit of an airplane, right? And so being able to pare that down to a digestible amount that you can share in a reasonable timeframe has been really the challenge.

And frankly, Epic and really all the other EHRs have been unable to do it successfully up until now. And now this generative AI can create this concise summary. That provides a clear understanding of what happened in the last eight hours with this patient. Key changes, key things that should be communicated to the next oncoming nurse.

That is huge.

at the different categories, right? AI is just, it's quite, it's a banded around term. I don't think there's a vendor that hasn't mentioned about their amazing AI capability suddenly in the last three months, right? From nowhere. For us, yeah, on the generative AI, I think there's amazing opportunities and to Reid's point with Epic.

I think they're doing well with their Microsoft relationships and leveraging some of certainly in box in basket, but then you've got the ambient listening, right? And more of the the voice technologies, which is super interesting. the interactions that it hears so that the doc can quickly move what's relevant out there into the patient record.

I think that's huge and something that, we're looking at heavily generative side though. Things like legal contracts for us. I think this is going to be like a game changer, right? Where we've traditionally had some of the legal Analysts constructing these lengthy documents.

If, and it's important that you're obviously using your own instance here. So we have our own instance of a private instance of BARD right now. And are starting to play with some of the generative capability there. We're pulling together, hey, I need a contract that is of, what you put in is what you get out there really.

But the 10 main points and bam, this thing is pulled together. together pretty well for a final human, run through and it just it's going to get better and better over time. So I'm excited. Some of the behind the office roles that were traditionally, let's say, white collar that I really think are going to be I won't say displaced.

I will say we can focus on, what did you say, Joybucket read? I like that. So the non Joybucket work we can move it elsewhere. So there's some really cool stuff occurring in that space.

It's interesting. in a whole bunch of the tools that I'm using, and again, I'm with a small startup.

I'm starting to see these boxes show up on it. Would you like generative AI to help you here and help you here? I'm curious because I was in a discussion with CIOs and they were talking about The various tools and the AI capabilities, especially the generative AI capabilities that are being layered on.

And the discussion quickly went to licensing and cost. we expect it to be a significant productivity gain, but if you start layering that on 10, 15 tools and all of a sudden it's an extra 20 per user per month, agree with what you guys are saying, which is, hey, just keep an eye on your existing vendors.

They're all working on it right now. They're all going to try to integrate it. Are you a little worried about the licensing?

No,

always. Yeah. If Microsoft are involved with chatt, I'm really concerned about my E three and the 25 pages I have to pick through every year. Yeah, there's gonna be a associated cost, right? And I think from the, but there's, from the containment standpoint, that's why I think it's also important that you're running your own instance so that potentially you can run some of your own compute over that to to keep cost down.

Think, Bill, on that, and let's just take Microsoft, so you consider their huge investment in open AI, and they're going to want to monetize on that. So there's going to be cost of this additional functionality. So we try and look at things where we can maybe have more of a fixed cost going into it.

Sometimes in Azure world, Microsoft There's some sticker shock, and you don't quite understand, how the pennies start to add up. But if you look at their co pilot model, where it's a fixed cost, it's a licensing entitlement, like annually, then it's a little bit easier to potentially calculate that ROI, return on value, based on the return productivity or gains there.

So that's how we're exploring it, is trying to really, as much as we can, have a tight kind of TCO about what we're doing going in, and then starting small, measuring that, like proof of concepts before we even pilot, and then not just scaling it to everyone in the system. But only entitling those licenses to users who are going to actually then glean that benefit and that value.

Because you're right, there are costs to this. There's no such thing as a free lunch. The one thing I remember from Econ 101.

  📍 We want to thank you for a wonderful year. As you know, we have celebrated our five year anniversary at This Week Health, and we are going to enter our sixth year of doing this. And we set out a goal to raise 50, 000 for childhood cancer this year, and you did not disappoint. We have raised close to 60, 000 this year for childhood cancer, and we really appreciate you.

We appreciate you. The community coming together. And we hope to do more of this next year. We hope that you'll join us. 📍

  📍 Yeah. Let's go ahead and Lee, this one's specifically for you. What's the immediate impact of AI on medical imaging? What's the thing that's being done today that's impacting medical imaging?

The immediate impact is turnaround times. So if you're part of a large system, you've got a queue of images to be read. If you're part of a system that does multi modality, so CT, MR, MAMO, DEXA, X ray, ultrasound then those queues can get quite large. I can tell you in my last place, the queue can get as high as 30, 000 studies, not images, studies to be read.

And that can take quite a while. And for a patient who has, a medical concern, Waiting an extra day or two for a result is really impactful for the patient. As AI comes into the workflow, and we can decrease the amount of time it takes to actually turn that around and get a result to the doctor and to the patient, that's really impactful.

In the case of x ray, for us within about two weeks, we decreased our turnaround times for long boat x ray by 36 hours. That was impactful.

That is impactful. There's a couple of questions, as you would imagine, on ROI. How do you present ROI? How are you thinking about ROI? AI conversations take me back to when we were having discussions bringing computers into organizations that didn't have them before.

almost that kind of conversation where they're like, Oh my gosh, look at all this extra cost we have. I can't believe we're now going to buy this and we're going to have a special room for computers and stuff like what, why this is all extra cost. What, where are we going to get it? And now we look back at that and laugh because we're like man, everybody was, is so much more productive in theory today than they were in a lot of industries.

They're a lot more productive today because of technology than they were.

but Bill, there was an impact to that, right? Like I think back on imaging specifically, I remember when we went from, silver based films, right? Actual plastic films that, we would hang or look at over a light.

To the digital transition. I was I'm dated enough to be able to say I was part of that. And I remember there was a whole group of radiologists across the country who retired. IT was just too much. It was an introduction of too much change into what they had known to be the case for 40 years.

And they just couldn't do it. And I suspect with AI, there will be a similar, maybe not quite as steep, but definitely there'll be a similar adjustment within the medical communities.

Bill, can I add on to that? Because I think, and it ties into a question there, around, ROI and how you're going to sell it to your boards, your executive teams and stuff.

For us, I think it's really easily quantifiable, but you you have to start out with a bucket of innovation money or research money. You have to have these things come along, these texts come along, so it shouldn't be you saying, Okay, chat GPT is out. Let's go and say, Hey, we think this could really change something for, no automation.

And then can I get some funding to go try? It's the other way around. You've got to be constantly reinventing and trying to displace some of the friction in your business as it exists. Then you can start to say, Hey, we think Through these tests we've done, here's the ROI, right? We're displacing so many manual workflows, especially through the application of OP8, which if you want, is really some of the original AI, right?

It's just like taking manual friction out of a process and automating it through compute. And I think... Every year I can show you how much we think that we're going to be able to save applying some of these techniques, but none of it would have ever happened if we didn't have some of that research and development money to go play and see what tool could liberate the business potentially.

So if I hear you right it's almost like you're going into the meetings. You're listening to. The, at St. Joe's, there would be 180 projects brought forward every year. And you just sit there and you listen to it and go, AI can be applied to that. It can be applied to that. It can be applied to that.

And it will change the way we're looking at it. And then taking a piece of that and saying, hey, I know you're going to go in a traditional route here, but can we try something out here and see if it, if this might change the way you think? Or can I bring somebody in who can introduce a tool that's, Doing this in a little different way.

Is that an approach that you're talking about or is it different? Maybe. And I think

you've seen it in action in our prior lives, right? It's more about, and it's still a technique that I use to this day, right? We have a small group that are responsible for some of this emerging tech and they see what's coming and they have the permission to play.

But they're sat in offices, right? It's where some of our original, back in the day logon stuff came in and stuff. Like you're sat there looking at the pain that the poor providers are going through is how they can't, something is frustrating them or slowing them down in their workflow.

That only came through sleeping in the, in the ED back in the day. And for us right now, we have those folks. That sit in those offices just watching where's the friction and then how can we apply some of the technology to the friction, not the other way around. So it's living and breathing and day in the life really.

And then through that, I think our solution comes.

Reid, how are you talking about ROI of these kinds of tools? Yeah,

we're, it was mentioned earlier by David, there's. Like AI is everywhere and it's infused now in vendor and marketing materials. And there's some of it, quite frankly, that's going to be vaporware.

So you can't apply a supply side kind of driven approach to AI, or you're going to drown and you're going to miss it. So we really start with the demand side kind of lens. And so really be clear going into it, not just, Hey, here's an AI tool. How do we apply it? But really start like back to Clayton Christensen.

What's the job to be done? And we found that if we're really specific about that, if we then add the other lens of our system strategy and focus areas, that then provides a pretty good kind of microscope to dial in on, okay, here's some AI capabilities that we want to try. And then from there, the ROI, and it's just...

Stuff we all do today already. Some of it is quantifiable, but some of it is also qualitative. So like provider satisfaction. We can do surveys, some of that's hard to measure. But we found that if we go into it with being able to explain the why, like here's the demand that's driving. the investment or this pilot or proof of concept it's a much easier conversation to then paint the picture about the hard kind of calculated cost ROI, but even maybe the softer return on value equation of it.

But it's an ongoing dialogue

for sure. All right. This is an interesting question. We've talked about generative AI. Person says, my executive business mentor was talking about generative AI and was saying essentially it's an essential skill set for almost every role going forward. They have to have some working knowledge about generative AI, at least have a functional level.

as a user experience, even if they are just getting started with it. I don't care what the role is in your organization. They've had plenty of time to get in and play with it. They either have a growth mindset or they don't. It's an essential training in today's day and age. Is it an essential training in today's day and age?

In my business, it absolutely is, but again, I'm technically in media and production, and my business is words. So generative AI is everywhere, and you're seeing it everywhere. But I'm curious if you feel the same way that will be a... What is the standard set of tools you want your staff to have across the board?

I love that question and I agree with it wholeheartedly. Yeah, this is the next generation of, tool sets that we need to be tooled for. And it's it's going to be a standard across the board. And I do think we should be aggressive with making sure that we're retooling as we go.

Otherwise, you're going to get left behind. And I think there are traditional healthcare. It's a hard one because it's slow moving to start and then there's going to be so many folks pointing out all of the risk and all of the danger that it's going to pigeonhole someone if they're not a strong technologist and not an advocate to move some of these tools along because they will remove friction and provide better patient outcomes.

Lee, I'm curious, quiet on that one.

Yeah, think we're at the front end of this. I was just pondering all the different roles in a hospital system or an imaging company. And, some of them are extremely manual and physical roles, right? So I'm trying to imagine a scenario where that would happen.

I see it this way, that it's going to be a continuum. And as time goes on, more and more of a percentage of the employees will need to have some working knowledge of AI. I also think it's incumbent upon the system, the company, to build in frameworks so that people can be successful at using the AI, right?

It can't just be go use the AI. It's got to be, here's our normal workflow, here's what you log into, here's the problem you're trying to solve, and here's how you leverage this particular tool to help solve that problem. It has to be guardrails on it, I think, in order for it to be effective.

Yeah, I think

I had an aha moment this summer and not even at the system level, just in our department was just giving a kind of a brief State of the Union update on AI and how we're thinking about it, and I had an erroneous working theory which was that, yeah, everyone in our department they're technologists they're excited about this, they're using it, and in the team's chat, as people are asking questions and having dialogue during my update, It became very clear that, and it's not a majority, but it's a bigger percentage than I would have thought.

There was voice concern about why are we doing this? This is unproven. This doesn't feel safe. Like we're healthcare, safety is the watchword. So I've had to dial my enthusiasm. I'm still an 11, but it really highlighted to me just the need to meet people where they are, to have the conversation, to listen and understand.

What their concern or hesitation is, to not make them feel forced, and to everything Li just said. Here's the guardrails we have in place to avoid the very risks that you're highlighting. Here's the perceived benefit of how, as you lean into this and learn how to use it, it's going to allow you to do things that make us safer.

But it's just, I think I overestimated just how readily people would accept this as what a great new tool that's going to help us be better. We've had to really spend some time listening and dialoguing around this.

It's interesting. This morning alone, I've written about 400 lines of code.

Of that code, I think I wrote 10 of those lines of code and ChatGPT wrote the rest of those lines of code. The code works great. Yeah. I'm much more productive, much quicker. I think generative AI. will do more for IT organizations that adopt it than any other area in the hospital. In the near term, I think it can be applied in so many various different ways.

And then the other thing is, quite frankly, your whole team doesn't have to understand it. The programmers do. And I think this is an opportunity to layer in programmers in between the the functional interface And what's coming out the other side. I know I'm supposed to be asking the questions.

I know you guys are looking at me like, are you talking? You should be asking questions. The reality is, I heard Satya Nadella talk about this, and he said, look, it's a natural language front end, it's a reasoning engine, and it's a co pilot design construct. Those are the three things that He emphasizes, and you've emphasized a bunch of those already, the copilot design construct keeps the human in the loop, keeps the human responsible for the output. Somebody's reviewing it. I'm looking at the code after it creates it and going, yeah, that's going to work, or that's not going to work. But it's that reasoning engine.

That we do a thousand times a day that is stuff that does not require a human being anymore. And, it's and that's the stuff I'm programming today and I'm just, I'm passing it through the generative AI engine and it's kicking out outputs. And I'm looking at it going, that's better than me hiring a person who's been in the industry for 10 years.

So yeah, I'm going to start using it. I'm not in healthcare. I'm in media

Just if I could quickly say to your point, I think for health systems, they have a choice to make around whether to build a development team or not, right? That's an expensive entity, right?

And so you're asking the question, what output can I expect from the investment of building a team of two or three people or 10 people or 20 people, whatever that might look like, and you're at, and you're doing that value equation in your head. What am I getting for what am I putting in, right?

What's happening now is the ability to get output. is Becoming a 10 times, compared to what it was before for the same input from before. So I think the equation itself is changing for health systems. And I think some systems that perhaps haven't had a team or had a smaller team may have the motivation now to double down on that.

I

agree. So who's setting the policy? or governance around the use of AI in your various organizations. Is that a concern? And Reid, I'll start with you because you probably have, if I'm guessing, probably most significant conversations around that.

Yeah, so we've set up what we call an AI Advisory Council Cross Functional Group.

We have there, obviously legal, compliance, DEI members there. Physicians, we have clinicians there as well. And it's been really interesting because one of the early on this topics was should we block these LLM tools in particular kind of out of the gate until get our arms around it?

And this group came back pretty emphatically with, we don't think that's the right thing to do. And it's really problematic because. With the each day there's going to be a new LLM tool that's out there. So the cat and mouse game, you just, you never win. It's going to convey the attitude that we're scared of this.

We don't want to use it, which is not true. So it's much more about education and just making people aware. And, but just every time we think, Hey, we've got this figured out, or we're getting close, there's another wrinkle. And so recently it came up that  what if we have a meeting and there's external parties?

And to use the name of a tool, they're running a tool called Otter AI. And it automatically, is the bot in the meeting and then transcribes the meeting and sends out notes to all the participants, totally outside of our control, to St. Luke's meeting. What's our response to that? How do we control that?

How do we create education about, that's not allowed? And if it does happen, what's our response? So it's very much a work in progress. And right now we're, we've landed on. We don't think we need a brand new AI policy. We think that this is an addendum or a tweaking to the existing acceptable use policy.

And we're just going through the effort to really dial that in.

Interesting. two are with different organizations. David, I think Pacific Dental is privately held and Lee, you just left a private equity backed organization. are the governance conversations a little different in those kinds of structures?

Yeah, that's a good question. Probably from my standpoint, I think we can be faster moving, but obviously we're, I think we specifically, it's why we have our innovation group, and then, our core business, and the core business, and a health care company has to have guardrails, right?

But we can move fast, right? And before we go outside in the testing part, and then when we move into more of operational deployment and use within the general business, then obviously, there does need to be those lanes. And I agree with Reid, it's more of an addendum. AI is like a modern day spell check is what it's going to end up being.

I think it's what everybody will, begin to use. How do we automate, this previously manual process, right? Essentially, and obviously there's more intelligence around the generative piece, but again, I still think that's also like a modern day spell check. So I I wouldn't let the FHIR police get too far over the top of it and stop the innovation.

It's, the important part for us around when you look at more of the, the ChatGPT and the BOD and those kind of, systems is It's making sure that you do have a private instance when you're playing with some of your corporate IP, if you like, because I think it's dangerous with what you could feed out there on the public market right now.

It's just learning huge amount and maybe some of your secret source shouldn't be out there. So that's the guidance we're giving people is don't. Load up, these huge detailed questions, because the power is in the questioning. We had a conversation about this recently, Bill where I was with the CEO of NVIDIA, and he was just talking about the enormous power and compound interest, if you like, in the questions being asked.

That's where the real learning was coming from on their side, not the ingestion of, the mass amounts of data. It was learning what do people want to see, and through the questions come the answers. So it's like a little bit of a mind bend for me.

I could just dovetail on both what David and Reid said, I think great AI is grounded in great data and great data is grounded in great data governance, right?

So if a system or a company has a framework in place for data governance around accountability of the ingest of data and a process for the development of policies and procedures around that, I think the AI piece dovetails really nicely with that. And that same team that's working on excellent data governance can be the same team you leverage for policies and procedures around AI.

hEre's an interesting question for y'all that came in. And it's essentially, how are you doing it? How do you stay current? How are you finding out what's next, what's new, what's happening? David, you have... team of people that's specifically tasked with stay ahead of what's happening, and I assume you go into that little bullpen and they tell you all the cool stuff that's going on and you probably just geek out for an afternoon, but is that primarily how you're staying up?

Or you mentioned being in the room with the CEO for NVIDIA. That's not a bad way to stay current on what's going on either.

Yeah, no, it's good. There's some amazing, networks that we stay connected with. Still many of them out up in The Valley, and I was at an event recently where Tim Cook was there, I got to meet him and the CEO of NVIDIA and, there are different opinions with AI, right?

Tim Cook was saying, we've been doing it for years. When you fall over with your watch on, for example, that's AI. When you get into a car accident, that's AI. We've been at it. We just call them features. He believes it's part of a static toolset. We're just in a hype cycle right now.

Whereas NVIDIA were like, this is it. This is the next industrial revolution. This is... But then, they're selling major compute and processes behind it. Depends on who you're asking, but yeah, our innovation team is set up specifically to test. We set aside a bunch of time for clinical device testing, like 3D printers and whatnot, which is a fast emerging product in the dental space, all the way through to different, software solutions.

There's no end of... of folks trying to sell you stuff. So it's just one, I think we're well connected in our networks with what's going on from the largest tech companies in the world. And then two, we like to test and play, and we encourage folks to fail fast.

And we may test, 10 products a month. And one of them could be a sure thing. The other nine will just go to the wayside. So it's just, getting through it and staying on top of it. We kind of love to play with it. The toys and a fortune that we're able to do

so I had the opportunity to walk through that in Southern California, but you essentially have an office set up like a, and I've seen a couple of labs for healthcare systems now that have essentially a mock up of rooms and various things so that people can bring in technology, roll it in, roll it out, different sensors, different, that kind of stuff.

I thought that was an interesting approach that the team is not only just sitting in their cubicles going, hey, this would be really interesting. They're walking around the corner, they're actually printing out the, not going to use the right terminology, but essentially printing out the tooth that is going to, be used the 3D printer.

And so I think that kind of, creating that kind of environment and setting aside some resources to stay ahead because it is coming at us pretty quickly. How, Reid, how are you staying ahead of it?

I think it's, if I can recall an earlier comment, it's just, this is a shared journey.

It's more about collaboration than competition. So it's staying close to our partners, understanding their roadmaps, learning from them. I get a lot of value from peer dialogue. In fact, just last week, Lee met with my team to just share some knowledge. of what he's done in the imaging space and what he sees in the future.

Bill not to suck up, but I think some of the interviews and the conversation you have been really good. You did one where you debriefed on the Noteworthy conference, and I gained some good insights there of how to think about it from an Epic landscape that I hadn't thought of before.

But I think, the underlying trait is humility. There's no expert in this space. And so we've just, we adopt an always learning, like more curious than certain attitude and that's served us well.

Cool. I think this is going to be the exit question. apologize. I think I got through a third of these questions.

But because there's so many questions around this right now, there's just so we could have a webinar every week on this every day, probably. Here's my exit question. It's gonna be very specific. And I want each of you to point to one area that you think you're going to be implementing AI in the next six months, like one area you think you're going to be doing it that you believe there's going to be an impact or a return for your organization. And Lee, you can pretend like you're still at Snyman Med. And Lee, we'll start with you. If there was one area you were going to implement AI and imaging, where would it be right now?

So in the imaging space there's a ratio known as the conversion factor. Okay. Which goes from the order, which comes in, which could be a paper order, God help us, a fax order or an EMR integration order that comes in. And then a patient gets scheduled. And then from that schedule the patient arrives and images are acquired.

So that whole continuum, that process, there's a conversion percentage you're trying to hit that. On the scheduling front, many places spend a, just an insane amount of money. On people to do this work, right? You get thousands and thousands of people in a contact center to be able to do that middle portion of the conversion, the scheduling itself.

That, in my opinion, in the imaging space is the most amenable right now to applying technology that automates. And creates a scenario where the human being involved has to do the minimal amount of work. For example, the AI could do, 90 percent of the information gathering that's required to actually make that appointment.

And maybe the individual the staff member scheduling has to do just a couple of things at the end to tweak it a certain way. Or maybe we get to the point where the entire thing is automated. There's a lot of really interesting technology in that space. From an ROI perspective, if the R is measured, not so much in patient satisfaction or physician satisfaction or staff satisfaction, but really in hard dollars.

In my opinion, there's a lot of bang for your buck right there.

Yeah, I'll agree with you. David, next six months where would you take AI, or where do you think you could have return?

We're heavily implementing the image in space right now. So that's been, that's our priority is finishing that off in all the offices so that we've got Essentially six of our major pathologies identified by tooth number in advance of the doc sent down with the patient chair side.

So we think there's massive power in that. We've been investing in that for a while, but also in the next six months, the The inbox and ambient listening is something that we'll be doubling down on and seeing what we can do and we're heavily implementing a lot of we're seeing through the application, more of the machine learning and outputs data wise, a ton of gaps that we're uncovering without the use of humans by appointing massive data sets in the RevCycle space.

I think it's like a goldmine of information. I'd encourage folks to, on here as a little takeaway, because I've been looking at it as well, and stuff's coming at me so fast. I took a ton of notes when I met with Jensen, the NVIDIA CEO again, and he's, I was just getting my bullet points, my notes, and he just said, you have to now.

Vectorized, tokenized, applied neural models. It stuck in my head and I was like let me, what does he mean by that? And I've been, doing a lot of research in it. But the vectorization, which has been around for a while, really starts to break down. It's essentially, here's information, the information then gets labeled, and that's how we begin to train these models.

And it's deep and super interesting. And I think, listening to these guys talking about vectorizing everything is really like the foundational pinnings of the outputs that you're going to get with AI. Go deep research in the space. It's a really fun time. I think like I said, in AI has been around for so long in medical and we're just catching up here.

It's an amazing new thing. We're just doing what various groups in medical have done for years and now just applying it to teeth. It is exciting for us, but that's not the cutting edge stuff. That's going to be the bread and butter stuff.

Yeah, so RevCycle is absolutely and yeah, I was just talking to somebody and they said, do you realize we've been using AI for well over a decade or so in healthcare?

And they started pointing out some things. I'm like, oh yeah, you're right. Reid, you get the last word.

Yeah, so for us, without a doubt, it's ambient listening. So I'll just double click on that for a minute. We've been talking about it for a while, but what's been really beneficial for us is the emergence again of generative AI.

It has brought new vendors into this market space. It's a better experience and product than it was a year ago, and at a fraction of the cost, which is incredible. But I've never seen anything that our providers have had more kind of amped up anticipation for. So we're just about to kick off our proof of concept.

And we talk about experience a lot, and what we like about this is Sometimes you decouple patient and provider and they're sometimes at odds because what improves the patient experience actually adds burden to the provider or vice versa, degrades the patient experience, but the provider's happy.

This is one of those areas where it can tick all the boxes. And also the ROI is there because it solves the primary access challenge where we can see more patients. So really interested in this and hoping that our working theory is proved out as we start to interact in a live patient environment.

Great.

Hey, I want to thank everybody who has been a part of this. I definitely want to thank the three of you for being here. You guys have all been great friends of the show and have great conversations with you all. All year long around this. For all the participants who've been a part of this.

Thank you very much. This will be out on our podcast channel probably I don't know. I make these commitments. Then my team looks at me like I'm in insane, I think tomorrow. Yeah I think it would be 30 days. It'll probably be out in December. We'll release it. Maybe we'll release it over our production break between Christmas and New Year's and get it out there for everybody.

Thanks again everybody for being a part of it. Thank you all.

Thanks. Thank you.   📍   📍 I love the chance to have these conversations. I think If I were a CIO today, I would have every team member listen to a show like this one. I believe it's conference level value every week. If you wanna support this week health, tell someone about our channels that would really benefit us. We have a mission of getting our content into as many hands as possible, and if you're listening to it, hopefully you find value and if you could tell somebody else about it, it helps us to achieve our mission. We have two channels. We have the conference channel, which you're listening. And this week, health Newsroom. Check them out today. You can find them wherever you listen to podcasts. Apple, Google, overcast. You get the picture. We are everywhere. We wanna thank our keynote partners, CDW, Rubrik, Sectra and Trellix, who invest in 📍 our mission to develop the next generation of health leaders. Thanks for listening. That's all for now.