Subscribe to The Podcast by KevinMD. Catch up on old episodes!

Join Adam Mariano, a health care executive. We’ll delve into the world of Big Data and tokenization in health care, exploring how these technologies are revolutionizing patient insights, data security, and health care outcomes. Learn about the latest advancements, practical applications, and the future of data tokenization in this dynamic industry.

Adam Mariano is a health care executive.

He discusses the KevinMD article, “Improving patient and population health insight through next-generation tokenization of real-world data.”

Our presenting sponsor is Nuance, a Microsoft company.

Together, Microsoft and Nuance are leveraging their rich digital technology and advanced AI capabilities to tackle some of health care’s biggest challenges. AI-driven technology promises to revolutionize patient and provider experiences with clinical documentation that writes itself.

The Nuance Dragon Ambient eXperience, or DAX for short, is a voice-enabled solution that automatically captures patient encounters securely and accurately at the point of care. DAX Copilot combines proven conversational and ambient AI with the most advanced generative AI in a mobile application that integrates directly with your existing workflows.

Physicians who use DAX have reported a 50 percent decrease in documentation time and a 70 percent reduction in feelings of burnout, and 85 percent of patients say their physician is more personable and conversational.

Discover AI-powered clinical documentation that writes itself. Visit https://nuance.com/daxinaction to see a 12-minute DAX Copilot demo.

VISIT SPONSOR → https://nuance.com/daxinaction

SUBSCRIBE TO THE PODCAST → https://kevinmd.com/podcast

RECOMMENDED BY KEVINMD → https://kevinmd.com/recommended

GET CME FOR THIS EPISODE → https://earnc.me/gLeNn7

Powered by CMEfy.

Transcript

Kevin Pho: Hi, and welcome to the show. Subscribe at KevinMD.com/podcast, and get CME for this episode by clicking on the CME link in the show notes. Today we welcome Adam Mariano. He’s a health care executive, and today’s KevinMD article is “Improving patient and population health insight through next-generation tokenization of real-world data.” Adam, welcome to the show.

Adam Mariano: Thanks, glad to be here, Kevin.

Kevin Pho: So let’s start by briefly sharing your story and journey to where you are today.

Adam Mariano: Oh, I have one of those odd resumes, pretty well into the health care ecosystem, the health care tech ecosystem. I actually started off as a nurse, was in the clinical space for quite some time. I’ve spent about 25, 26 years now in, I’d call it health care IT, so analytics, warehousing, software development, integration, big data journeys, cloud strategy, master data management, we can go on and on and on. But I also happen to be an attorney with a focus in civil rights, and I have a master’s in population health, so I have one of those odd arcs.

I spent a bunch of years in the clinical space, and I got into the transition to IT and software development in the life sciences space, and slowly moved into warehousing and analytics, and spent the better part of 15 years or so consulting with payers, providers, life sciences, and retail health organizations, which used to be pharmacy. Helping them figure out how to do a bunch of things. One, provide real value with data. Two, get organized around a data-driven principle. And then three, over the last 10 years or so, really figure out how to effectively develop cloud-smart organizational strategies and taking advantage of hybrid environments.

But all of that has been with the disruptive focus of trying really to drive equity in health care. And so the civil rights attorney in me will not die, and I have, I guess, some sort of unending sense of hope, in that we will continue to push that ball uphill and continue to move some of that journey forward. So that’s the short story anyway, that’s the elevator pitch.

Kevin Pho: All right, so let’s get into your KevinMD article. It’s titled “Improving patient and population health insight through next-generation tokenization of real-world data.” Now, give us a brief definition of data tokenization as it relates to health care, and then use that definition to lead into your KevinMD article.

Adam Mariano: Well, generally tokenization in the health care space is a way for health care organizations, research institutions, life science companies, and the like to leverage research data in two fashions. One, you can use tokens to link data sets that are identified, so making it easier to leverage that data but not having to ship it to multiple processes, so you can think of it as a unifying backbone. But the other way is to share data into a research ecosystem that is deidentified but is all linked through the same unique identifiers. You can think of it as a golden record.

So I think in the article I use the example of my history as a child. My name changed when I was in the sixth grade, so there’s two versions of me, pre-Mariano, which was Williams, and post-Mariano. And you can imagine that my record is bifurcated. But you see the same thing in people who get married and change their maiden names, you see children who are adopted, but you also see folks who go through all sorts of change in their historical alignment to the health care institutions. So the payer they were under, the health system that they got their general care from, a new specialist they got, that claims data, that socioeconomic data, the EMR data, are all linked by different IDs.

And tokenization allows us to link that backbone by sharing what we refer to as hashes, which are essentially subsets of that data that in their own right can’t identify a patient, but when linked together give you a backbone that allows you to do that. And so in general with tokenization processes, the nice part is none of those institutions actually have to ship PHI outside of their four walls. The tokenization engine tends to live behind their firewall, only ships out hashes, that all gets linked by an independent third party, and then they can start to use that key to link other data sets. So really an exciting innovation from a health care perspective. It’s been around for a while, but it’s been slowly evolving over the last 20 or 30 years to be something that’s really more useful.

Kevin Pho: All right, so now with that definition in mind, tell us some of the practical applications that one can see from tokenization.

Adam Mariano: Well, pragmatically, almost every drug that’s ever been prescribed has been built on the back of some sort of tokenization effort. So either in-study or post-study from a clinical trial, or even a post-market safety study, somewhere somebody is trying to link claims data, EMR data, other outcomes data back into the research data set. That is almost always done through a tokenization effort. So those data sets are shipped through a tokenization process, those links are all used, and then that deidentified data set that the manufacturer gets to use for their research is linked up, and really enhances obviously not only the study data set but then also the outcomes for the FDA and for validation.

On a more operational footprint, if you think about a large health system that was trying to drive either a health equity program, an outcomes-based research program, trying to do decision support for physicians in care management, pop health, it’s a very easy way to link some of that operating or performance data, as we call it, across the backbone, without having to ship it all around to all of those applications and have it all be embedded. You can just link some of that data and use it for a broader research case. It also allows you to bring in third-party data sources in a much easier fashion. It shortens the onboarding time, it shortens the total cost of ownership. There’s a lot of really, really good reasons to use it.

Kevin Pho: So you mentioned that data tokenization can also be used in addressing health care disparities, certainly addressing social determinants of health. So talk more about that.

Adam Mariano: Well, I think we all know, as you look across the broad research landscape, the carve-outs that are missing are minorities, low-income patients, and hard-to-reach patients who don’t have access to tier one research institutions. And so as we start to expand clinical research, and we also start to look at driving more operational efficiency for patients at risk, the reality is you can bring in this tokenized data at large scale, which you would not be able to do in individual collection.

So for argument’s sake, you think about all the community health centers, which are sort of the linchpin for providing care to low-income, high-risk patients, lots of Medicare Advantage patients who are dual. They’re not tier one research institutions in a lot of cases, a lot of times they’re federally qualified clinics. They don’t have the capacity, the capability, to link all of that data, but they can leverage a large third-party data set that actually gives them some understanding of the served ecosystem and geography that they’re serving, the relative risk that those patients are under, deflections in those patients, we’ll call it their consumer lives for lack of a better term.

Somebody becomes unemployed, somebody used to live close to a transit center that now closed, there’s no more grocery stores. Which happened to me when I lived in West Philly and was a young parent, that the closest grocery store was a mile away, and then that one closed, and then you didn’t have anything in the two-mile radius. And if you don’t have your own transportation, you’re not on a transit line, that’s a big implication.

So what this allows us to do is really to bring that data in and link it to things that happen at a zip code level, but also on large data sets, where, for argument’s sake, at LexisNexis where I’m a GM and president, we sell a social determinants data set that actually allows you to profile patients at that individual patient level and link their EMR data, their claims data, and their social determinants data to give you that full picture.

So all of the things that we lose sight of when patients leave our institutions and go out into the real world post-discharge on a complex surgery. If you knew that they were struggling financially, or that they crossed a threshold and were now eligible for Medicaid or CHIP, you might create some extra touches to help them. So it’s additional support. You think about all the decision support places where we go blind when patients leave and we don’t have that visibility, that fourth wall, if you will. It really offers us an opportunity to catch patients in what I like to think of as a social safety net.

That’s one good example. There are plenty of others. Obviously being able to expand the apron of clinical research and start to include patients who can maybe take visits at a far-reaching institution or at a small research hub or at the community health center that’s not necessarily inside of the standard research grid, and provide them with a low-lift technology solution to integrate that data. It not only allows them to figure out who might be eligible for a study, but what patients fit the bill from a clinical and a willingness perspective, but also about how you might have to adapt studies to make access for those studies available to those patients. So folks who have transportation problems or mobility issues or even cost issues with getting to appointments and in process. So a big landscape to open. I think there’s a lot of those sort of non-traditional use cases that we also look at, to see if we can’t help lower the burden for institutions that may not have the technological capability.

Kevin Pho: So data tokenization is often associated with privacy protection. So talk about safeguarding sensitive patient information in the context of data tokenization. Do patients have anything to worry about, about their protected health information revealed?

Adam Mariano: I think if you look at some of the genetic testing companies who just had a recent breach, I got a letter, I’m actually not a customer but I got a letter, your data may have been breached. We all should worry about safety. It should be actually tantamount to everything we do. We think of it as a tier one priority. Tokenization is all about data safety and security.

The trick to tokenization, which feels like a little bit of magic, is PII and PHI never leave the institution’s four walls. So in order to tokenize data with a third party, you actually only ship hashes. You can think of it as encrypted data that’s a subset of the data. You can never actually take that hash and get back to the original patient. It’s just data that makes up almost like a code, and you can think of it as subsets, recipes is what we refer to them as, and they’re essentially a way to match different versions of a patient. So with all of my addresses and my phone numbers and my one name change, you’re going to need a multitude of versions of Adam in order to link Adam’s data across time.

But the nice part is this little bit of software that does tokenization actually goes to the health system, sits behind the firewall, and only ever ships out encrypted, unidentifiable data for matching. So what it does is it makes sure that not only can that patient data never leave the institution, even if the data that got outside were to ever be breached, it’s not linked in any real way back to the EMR records or back to any of the underlying data, which is clearly a fantastic step forward.

The other big win is, with most matching processes, especially if you take sort of the HIPAA Safe Harbor pathway, it tends to eliminate a lot of the identifiers you actually need to match patients, and it scrubs a lot of that data, so you end up with some of the abysmal match rates we’ve tolerated in the life sciences community for a really long time, where you were used to only matching 15, 20 percent of the patients in a study to their external data. Which obviously, if you look at a full data set for researching the outcomes of drug development, you don’t want to look at a 20 percent sample, it’s a little scary.

So I think there’s real opportunity to bring this data together, really provide a reduction in risk, and improve the ability for a lot of low-income communities to participate, while also not taking new risks with their health care data. Because the last thing we want is patients who are already at risk, already struggling to get care, already on the short end of the stick from a delivery and a biased perspective, and certainly struggling from a lack of equity, to also then have a security risk on top of that. So we take that very seriously. It’s one of the great things about being in this service business, is we really get to help organizations do this kind of research and integration and growth without actually taking risks with any of that data.

Kevin Pho: We’re talking to Adam Mariano. He’s a health care executive. Today’s KevinMD article is “Improving patient and population health insight through next-generation tokenization of real-world data.” Adam, what do we have to look forward to when it comes to health care data tokenization in the coming year or so?

Adam Mariano: Well, I think one of the great things, and I hate to jump on the gen AI and AI/ML bandwagon, but if you look at the arc from the beginning of 2023 to the end of 2023, and there’s probably a thousand articles that will say what I’m going to say right now, they won’t follow it up with the next statement, but they’ll say this: gen AI and ChatGPT were a toy at the beginning of 2023, and by the end of the year every company with some skin in the game, including ours, was already investing in real ways to apply it, whether it was back office or insights or automation or what have you.

I think as we go forward, the idea of tokenization and the reduced risk associated with sharing data in this fashion is a way we’re going to control making sure that the AI uplift does two things. So one, it doesn’t create new exposure for risk, which none of us can tolerate. But two, it becomes more democratized and actually offers opportunity, as we talked about, for at-risk and underserved populations to participate in this next generation of care.

It also is unlike a lot of the other sources which are now going to get tied up in a lot of what sounds like copyright and trademark infringement around article sourcing and a lot of other things. People have borrowed a lot of content from the internet to build out their models. It allows you to make sure that you have original data that is tied to a critical process that you don’t have to leverage from third parties, and so you reduce that risk around any of the conflicts around data licensing and ownership. This is data from patients for patients.

And it starts to lead into that space where you can think about consumer decision support tools that actually take their own health record into account and can compare it to other patients like them. Places where we can maybe do some better early detection in high-risk patients when they leave, so doing some early pre-discharge planning, looking at patients post-discharge and creating an early warning system. There’s just so many places where we can extend our ability to help patients in the windows where we don’t have visibility what’s happening in their lives, to start to improve our next best action with them. So I’m excited about that, and putting more decision-making capability and education in patients’ hands.

And I think just another place for this is to be able to push content down to them based on their records. So if you can safely tokenize that data, you can safely get it into a model, you can actually start to turn stuff back to them about their newly diagnosed condition of diabetes, and, how can I help you be healthy in your life with this new thing that you’ve just taken on as a burden? So lots of opportunity, really exciting space. I think, to the point I made, it’s really deeply rooted in how do we do this safely and effectively, but also scale it. And being able to automate things like this really offers us opportunity to scale brick by brick pretty quickly.

Kevin Pho: And my final question, Adam, tell us your take-home messages that you’d like to leave with the KevinMD audience.

Adam Mariano: Two things. One, we are all responsible for health equity. So anytime I can help, but certainly our organization can help you build a case for health equity in your ecosystem, reach out and share your stories. And then two, tokenization isn’t just for life sciences. It’s sort of like, cereal’s not just for breakfast. You can really use it everywhere. So think holistically as you go forward, whether it’s with a vendor like us or somebody else, that it really does lower your lift operationally, and there’s plenty of opportunity there.

Kevin Pho: Adam, thank you so much for sharing your time and insight, and thanks again for coming on the show.

Adam Mariano: Thank you, everyone. It’s good to be here.