Todd Gottula | Generative AI: The New Frontier in Healthcare Technology

Healthcare Unbound_Todd Gottula: Audio automatically transcribed by Sonix

Healthcare Unbound_Todd Gottula: this mp3 audio file was automatically transcribed by Sonix with the best speech-to-text algorithms. This transcript may contain errors.

Healthcare Unbound Intro/Outro:
Welcome to Healthcare Unbound, a podcast powered by Clarify Health, where healthcare's changemakers discuss ways to advance care outcomes, cost, and affordability.

Saul Marquez:
Hey everybody, and welcome back to the Healthcare Unbound podcast. I'm super excited to kick off this episode of Healthcare Unbound with an extraordinary person. We're going to be discussing what AI is and the power of it in healthcare. It's affecting all of our lives, and our guest today is Todd Gottula. He is the co-founder and president of Clarify Health. He's a visionary technology leader with over 20 years of achievement, driving growth, innovation, and profitability in the high-tech sector. Todd strongly believes that deploying technology and clinical expertise can improve the lives of patients and those who care for them. It's this innovative approach that drives Todd to deliver scalable technology, and that's what inspired him to co-found Clarify Health with Dr. Jean Drouin. And so, with that intro, Todd, I want to welcome you to today's podcast and really want to thank you for joining us.

Todd Gottula:
Well, thank you very much for having me. I'm really excited about today's conversation.

Saul Marquez:
As am I, and so you've spent a lot of time diving into the topic of AI, generative AI. It's quickly gaining steam across all sectors. So really excited about this discussion with you. What does the future of AI in healthcare look like?

Todd Gottula:
Well, that's not just the million-dollar question, that's like the trillion-dollar question that I think a lot of people are discussing and very appropriately. So maybe the best way to dive into this topic would be to first set some context, and like, a lot of this is going to be some basics for some of our listeners, but in case we're talking to some folks who are not deeply immersed in the healthcare industry. I thought it'd be good to first ground on what is it that we're trying to do at Clarify and what are a lot of organizations seeking to do as we address what people often refer to as the waste in healthcare or the misappropriation, misdirection of resources in the healthcare ecosystem? So basics are that like, who pays for American healthcare? Well, it's you and me through multiple pathways, right? So our employers are taking dollars and working with insurance companies to ensure that we have access to the care that we need. Similarly, we contribute taxes as well as through other means to support the government-sponsored portion, forms of healthcare in this country, Medicare, Medicaid, and so forth. So what is it that we seek? Well, we want to make sure that those dollars are being used most effectively and efficiently, and we want to make sure that the healthcare that individuals need and deserve is most appropriately allocated to them. And so we at Clarify, since inception, have been seeking to deliver on those two goals, which is how do you ensure that individuals get access to the most appropriate care in the most efficient way? And there are lots of techniques to do that, but that's sort of where the new technologies and the emerging technologies of generative AI and other forms of that large language models and so forth are really unlocking what I sort of refer to as the missing Rosetta Stone of how to take an incredibly complex system and decode it in ways that we can provide guidance to those that are either delivering healthcare, supporting those that are delivering frontline healthcare, or allocating the resources to do so in that way that we've all been seeking, which is individuals getting timely access to appropriate care in an efficient means. So that, for me, is the real exciting frontier that we're on right now, which is that these emerging technologies are finally going to allow us to deliver on the promise in a much more fulsome and complete way than we ever have been able to date.

Saul Marquez:
Yeah, thanks so much for that, Todd. And really, some of the themes that you highlighted is right care, right place, efficiently done. And just knowing what I know about you guys, Clarify Health, I mean, with the data sets that you have, the time is now to really add value. So for everybody listening, tell us a little bit about how Clarify Health is leveraging this technology.

Todd Gottula:
I appreciate that, Saul, absolutely. So the initial premise of the organization, when we founded it eight years ago, was to take the growing available data assets that were coming onto the market as sort of an exhaust of the processing that organizations have to do, whether that's payers processing claims and adjudicating and making payments out to care providers or whether it's the service providers to doctors, hospital systems or also payers. They were making data available and that de-identified, fully HIPAA-compliant way, to organizations like us that we could then amass large assets to be able to track longitudinal de-identified patient journeys for the purpose of identifying where there was positive and negative variation in the system. And that's the fundamental underpinning of any good optimization solution, which is, first you have to understand what the historical utilization and outcomes of any services. So you look across any industry, good baselining and benchmarking is the cornerstone of then being able to drive improvement. And in healthcare, having the large underlying data set to do that baselining and benchmarking didn't exist. Well, it does now, and that's what we've created. Once you have that foundation, then it's a matter of digging in and identifying the patterns of care that lead to better or more appropriate, more efficient utilization of services and, therefore outcomes. And since inception, we've been using data science techniques, so traditional machine learning, random forest modeling, to be able to create case-mix adjustment that allows us to identify where there's, in specific measures, where there's variation. And essentially, the question that we're trying to answer is very similar to wins above replacement in baseball, which is absent the treatment effect of the physician themselves. So if an average physician had treated this patient population, what would we have expected the utilization and outcomes to be? And then you can layer in the actual care provider themselves, and you can say, okay, well, look, the modeling is saying that this physician reduced utilization of nursing facility stays in total joint replacements by two basis points or by two percentage points relative to benchmark. So this is the core of what we've been doing, but then, to be able to really understand higher-order patterns is what these new technologies are allowing us to do. I can talk about that in a moment, and that's what we've been developing, is saying, look, we've created this incredible baseline asset, we've created both observed and synthetic benchmarks on hundreds of millions of individuals and billions of patient journeys. Now, what can we unlock, that we've got that baseline asset and these emerging technologies for this industry at large?

Saul Marquez:
Yeah, and it's that database of information that you guys have that really is a true differentiator. So let's talk about applications, Todd. You know, there's so many ways that we could use this generative AI, right? What benefits and what applications does it provide healthcare analysts?

Todd Gottula:
So I spoke earlier of the sort of Rosetta Stone, as I call it. What I mean by that is the true linkage between a utilization of a specific service or set of service in a patient journey for a cohort and the resulting true cost to the system. So what is it ultimately that a payer is going to pay for the decisions that are made during a care journey? What is it that a provider is going to get paid? How are they going to be rewarded for the work that they're doing? And ultimately, what is the patient going to pay, whether it's through deductibles, copays, or coinsurance? That lexicon doesn't exist in an incredibly transparent way. Yes, it's true that we know for a specific payer, given the No Surprises Act and the price transparency data, what they will pay to a provider for a specific service code, but that's in a fee-for-service regime, and that's not linked to all of the independent and dependent events across an entire journey of care. And when I say journey of care, I mean, for example, something simple like total joint replacement. So from the beginning of the decision to operate through to successful recovery could also be an oncology journey where someone is diagnosed with a form of cancer and they go through a multi-year journey of treatment and care. Inside each of those journeys, there's many, many decisions, tens if not hundreds of decisions that have these sort of cost and outcome decisions that need to be made. And what we're missing is that sort of holy grail-Rosetta Stone that allows us to understand, if I make this decision here, these are the likely and evidenced cost and outcome impacts of that decision. What the technology that is emerging today is going to allow us to do, and I'm highly confident of this, is finally create that translator, that lexicon that will empower clinicians to be able to say, for this patient cohort, I have two decisions or a decision node here which with two potential pathways and I can be much more informed about what the outcome and cost decisions are of both of them. That's what this industry has been seeking for decades, and finally, with the data assets that organizations like Clarify have created and generative AI, we will finally be able to answer those questions, and it's an incredibly exciting time.

Saul Marquez:
It certainly is, and we talk a lot about behavioral incentives on the podcast and the push to value-based care. Is this what we need? Is this what's going to get us there?

Todd Gottula:
It absolutely is, because we talk a lot about the ills of fee-for-service, and I try not to go down that path. Instead, what I like to talk about is rewarding and incentivizing care providers, just like every other industry, rewards and incentivizes those that outperform or perform in the way in which we're asking or expecting them to perform, and in a fee-for-service environment, you get paid for activity, not utilization and outcome. Well, why have we struggled? It's because this lexicon doesn't exist to be able to definitively say, with peer-reviewed support for the work, that, this decision is more beneficial and more efficient to a patient cohort than this other decision. And so therefore, I'm going to reward you and incentivize you for making that supported, informed, and beneficial set of decisions through a patient journey. And it really is taking the great work that clinicians are doing and supporting it with the evidence that we finally have available to us at massive scale across specialties and across patient cohorts in those specialties, and then setting up an incentive structure that moves utilization payments to behavioral and reward structures.

Saul Marquez:
Yeah, and also, very likely, helps a payer quickly incentivize, like incentivize on the spot because you know that the outcome is going to be there, so therefore, you get that type of behavior.

Todd Gottula:
That is absolutely correct. Life sciences companies have been working with real-world evidence databases for a long, long time, right? Which is understanding when you apply this therapy, what do we see the observed impact of that therapy to be over certain time horizons. That same concept, that same notion, is now able to be applied to all other forms of care, and so you're exactly right. Once you have that, you can then move to incentivize, reward, reimburse at the point of the decision because you've got all of the real-world evidence that says, well, look, when the appropriate patient who's having a total joint replacement has their surgery at an ambulatory surgical center as opposed to an acute care hospital, we know that that's going to be significantly less expensive for both the patient and the payer. And again, for the right patient cohort, that individual is going to have less likelihood of a readmission, less likelihood of an all-cause readmission due to infection, and therefore better outcomes. We've got that repository, and so now we can make that recommendation and incentivize it at the point of decision.

Saul Marquez:
That's fantastic.

Todd Gottula:
That is going to be replicated over specialty areas and primary care because now we can unlock the power of these data sets with the technology that is now becoming more available.

Saul Marquez:
Thank you for that, Todd. Yeah, that's impressive, and we sort of, in the healthcare industry tend to want to see things and watch them evolve before we adopt them. Generative AI and AI, in general, is this one we should sit on the sidelines and see how it plays out, or do we need to roll up our sleeves and get involved?

Todd Gottula:
So, I love the question because you're absolutely right. That has been an observation that I have had of the healthcare industry, which is striking because I come from the financial service industry, which has sort of the complete opposite approach, which is that let's try everything as soon as it becomes available. I almost would answer that question by saying, what is on the horizon is going to be unavoidable, and it's going to almost be adopted because people aren't going to notice it. And what I specifically mean by that is that what generative AI unlocks is not just what we've been talking about here, about the power of being able to find patterns, multivariable patterns of optimization in large data sets, but large language models also put a conversational interface on top of it, and it's that conversational interface which is going to break down some of the barriers, at least I think, in healthcare. Meaning that because it won't be about, hey, you've got to log into this system, you have to learn this complex user interface to then be able to interrogate this underlying data set that's going to give you some observations that you then have to figure out how they apply to your practice pattern, which has been the historical model in healthcare. It's going to instead be, ask me a question, and that question is going to be presented back in conversational language. And so that whole user experience paradigm shift is going to be just fascinating to watch on its own, but I also am highly confident it's going to break down some of this resistance to adopting this new technology and the new insights that are going to be available because people aren't going to notice. They're literally just going to start interrogating datasets everywhere in their life through Q&A, you know, almost like fact-like interfacing. And all of a sudden, new information is going to be available for those that are out there on the front lines, and then you back it up with reward structures based on having a conversation. And I think this industry is going to shift in a way in which we haven't even begun to anticipate how quickly it's going to shift.

Saul Marquez:
Wow, yeah, it's certainly an exciting time. And by the way, Todd, I did want to ask you, Clarify Health recently announced a new release, right? Tell us about that, and for the listeners that aren't aware, Clara.

Todd Gottula:
Yeah, so Clarify Clara is the conversational interface that unlocks the power that sits inside Clarify's platform. So that data asset and the technology that we've built on top of it to unlock the insights and the pattern recognition, that is what we call Clarify Atlas, and Clara essentially allows users to read that atlas and understand where they need to go next or where they can go next. And so the initial release actually sits on top of the No Surprises Act data, the price transparency data that we have collected from all of the payers and providers that have made that data available. And so it allows users to interrogate literally petabytes of data to be able to understand where there's opportunities for them to make better decisions based on essentially how either they are going to be paid for the services they render or how payers are going to pay those that are providing care. And so it's essentially providing, again, that conversational interface on top of a massive data set, and it's not just, hey, what's the reimbursement rate for a DRG-470? It's, where are there opportunities for me to make better care decisions based on how I'm going to get paid? Now, it is also in the context of providers deserve to get paid for the great work that they're doing. So it does allow them to ask questions like, well, how do I optimize? What's the best pathway here, both for outcomes and for me to be rewarded? So it's an incredibly exciting first foray into adding this conversational, optimization-focused user experience that breaks the paradigm of, oh, I need to go in and build a query, have an analyst analyze the data, give me a summary set of results, and then I have to figure out what to do.

Saul Marquez:
It totally automates it. And so super exciting, by the way, folks, if you have been racking your brain around optimization efforts, quality efforts, ways to get profitable again, because let's face it, we've been dealt a difficult hand, especially providers right now working your way out of the COVID hangover, what can you do to optimize your practice, your organization? Clara might have some solutions for you. And so, Todd, the question then is like, what would a CEO of a hospital or a physician ask Clara? Like, give me an example of what I could type in and, and what I could expect.

Todd Gottula:
Absolutely. So where we are today is supporting use cases. Like, I'm, the contracting manager at a physician group, so maybe I've got ten docs, and I'm going into rate negotiations with my payer. And so what Clara allows that individual to do is to interrogate this national data set of what organizations have been paid for similar services by this payer in both my geography, as well as nationally, and how do my outcomes relate to how others are being paid. So essentially, the position that a physician group is looking to take is to say, look, I'm delivering better outcomes, therefore, I should be able to earn top decile payment for the services that I'm rendering, and Clara gives the user the ability to understand and validate that, know what others are being paid, know how their performance compares to others, and be able to create a synthesis of the specific service codes as part of that book of business that should be negotiated and what they should be negotiated to be. So that's on the provider side. And then, on the payer side, it's sort of the example of I'm going into a geography, I'm trying to understand what the existing service providers are in this market, how they're being paid by those that are already paying them for services, the other payers. And as I go into contract with them, how should I set my rate structure? And again, it's not about minimizing payment, it's about actually doing what I just said on behalf of the providers, which is setting up a tiered rate card to ensure that the highest performing providers in a geography are being rewarded for the work that they're doing.

Saul Marquez:
Well, I can't tell you, Todd, how much I love this, and the opportunity is huge. And AI will not take over organizations, it just won't, but organizations utilizing AI will take over the industry. And so, Todd, I hear your message loud and clear, we've got to adapt. And we don't have all the answers, so if people aren't sure how to engage with this technology and want help, what advice would you give them?

Todd Gottula:
So I absolutely would recommend everyone. Even, I said this to my parents. I said this to my brother, who's a high school English teacher, that you need to start playing with tools like Bard, getting a ChatGPT Plus account, it's $20 a month, and literally just start asking questions and getting yourself familiar with what is on the horizon, because this isn't just about healthcare. Every industry is going to be disrupted in a positive way because we're just going to change the way that we work with data, that we work with information, the ability, again, to have a conversational interface that is context-remembering, and context-sensitive. You just got to get used to it. It's one of those things where, it's like before the spreadsheet, what did people do right? And now spreadsheets are just part of life, and you're familiar, everyone is familiar with working with data. Email, same thing, right? It's just part of the way in which people interact with each other. This is one of those, and I'm not the only person saying this, right? Everyone's talking about how this is one of those things. So first step, just get familiar with it, and the tools are available, they're free, and you'll be, I think, amazed at how both easy they are to use and the underlying power that you can unlock in your own life. So that's step one. And then the second thing is, start thinking about the problems in your life that you have. Like, I wish I could answer this. I wish this is like this complex question that has multiple dimensions to it, and it feels like an optimization problem. At the end of the day, most optimization problems are pattern recognition and application-type problems, so start cataloging those because we're on the horizon of being able to answer most, if not all, of those types of questions, and that's the exciting frontier. And then, start asking the partners that you work with if they can answer them for you. Start making those challenge statements out to your own organization or the vendors and partners that you work with, because that's the best way to be in control of your own destiny here, which is instead of waiting for people like Clarify to come to the table and say, hey, this is what we solved for you, we want people coming to us and saying, hey, these are the questions that I'd love for you to the answer, because that just fuels us to start that virtuous cycle of partnership with our industry partners to be able to drive change.

Saul Marquez:
Wow, Todd, you have no idea how excited I am right now. And by the way, I love that you started with the personal life, like ChatGPT, because this technology is permeating every aspect of life and not just healthcare. So I love that you started there as sort of something we can all do. So I guess the question for everybody listening is, have you explored ChatGPT? Have you explored some of these other voice AI technologies? If you haven't, take that baby step, and then when you start doing that, you start to see the possibilities that this technology can offer and then say, what can I do for this optimization project? For example, I love that you started there, Todd, it's such a friendly way to get in, and so, super exciting to hear that you guys are taking this to the next level. And you know, the other thing is like, I get so excited when I jump into, like I was doing a search, I just got StarLink, and I was.

Todd Gottula:
I'm on StarLink right now.

Saul Marquez:
Oh, are you? Same here. So I'm trying to figure out the dual-band. How do I get a 2.4? Because I needed to connect this printer, so I Google it, and then I go to a Reddit feed that has an iterative AI in it, and I'm like, oh, this is so cool. And then get into HubSpot, and I see that they have a content tool. It's happening all around this, folks. This is the time. And so, Todd, I can't thank you enough. Like I'm leaving this conversation energized, and I hope everybody else listening is. What closing thoughts would you have for the listeners? What would you leave them with as we conclude here?

Todd Gottula:
So what I would say is, when these technologies, generative AI technologies, and others started to come onto the marketplace, there was an appropriate degree of skepticism and some people even degrees of fear. And what I would encourage everybody to do is take that step to get personally familiar with them, because it's an amazing opportunity, but it's one that has to be applied thoughtfully, and that thoughtful consideration is going to be driven by every single one of us. And so the more each and every one of us can be informed about the opportunity, the limitations, the boundaries, but also the excitement that we're all feeling about the problems that this set of technologies in each individual's industries can solve, the better off we're all going to be. This is one of those times in life where we all have to be involved to drive the landing place or the stepping stone or milestones of the adoption of these technologies. And I share your excitement that we are on the cusp of something that is truly going to revolutionize many industries, including healthcare, and I'm just really excited to be a part of it.

Saul Marquez:
Wow, yeah, and I totally agree, Todd. I love today's episode. So folks, this is one that I would hit rewind. Listen to it again during your run or your drive to the office. How can people find out more about AI and the things that Clarify are doing?

Todd Gottula:
So about AI, I mean, there's a lot of emerging podcasts and literature out there, but again, I would go back to what I said before, which is, like, if you don't want to sign up for a ChatGPT account, use Bard. I mean, Bard is incredible, and playing around with it yourself, I find to be the most productive way to get immersed because most people are experiential learners, and so digging in really would be a first step that I would recommend for everybody. And then specific to Clarify, I have a blog that we maintain and have been putting out some content on where we put out our first press release about Clara, which we're incredibly excited about. There's going to be some major announcements coming on the heels of that about some of the insights that we've unlocked using Clara ourselves against the data sets that we've pulled together that we're going to be coming out with as part of the Clarify Health Institute, which is our research arm that we put out publications through. So sort of stay tuned, watch this space, because there's going to be some really exciting evolution here over the course of the next 3 to 6 months.

Saul Marquez:
Amazing, Todd, thank you. And folks, just to reference that blog that Todd mentioned, look, Todd's a thought leader in this space, if you're looking to plant seeds in the garden of your mind as it relates to this, definitely follow Todd. We're going to link up that blog that he just wrote so that you guys could all check that out in the show notes. So make sure you go to the show notes, click on that, and remember to subscribe to the podcast because that's how you get all of the most up-to-date episodes to stay on top of technology like this. Todd, I can't thank you enough for spending time with us. Looking forward to connecting with you again soon.

Todd Gottula:
Great, thank you very much. Take care.

Healthcare Unbound Intro/Outro:
Thank you for listening to Healthcare Unbound. We hope today's episode was insightful. If you want more information on how Clarify Health can help you, please visit ClarifyHealth.com.

Sonix is the world’s most advanced automated transcription, translation, and subtitling platform. Fast, accurate, and affordable.

Automatically convert your mp3 files to text (txt file), Microsoft Word (docx file), and SubRip Subtitle (srt file) in minutes.

Sonix has many features that you'd love including enterprise-grade admin tools, generate automated summaries powered by AI, powerful integrations and APIs, transcribe multiple languages, and easily transcribe your Zoom meetings. Try Sonix for free today.