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 we have another interview in action from the conferences that just happened down here in Miami and Orlando. 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. We wanna thank our show sponsors who are investing in developing the next generation of health leaders, Gordian Dynamics, Quill Health, tau Site Nuance, Canaan Medical and Current Health.
ight, so here we are at HIMSS:I look forward to connecting with you. We're gonna talk. . Uh, you know, fire Interoperably, we'll talk a bunch of those things. Yeah. Last time I talked to you, you were in the administration. Yep. You since moved on. What, what are you doing now? I'm the Chief Strategy Officer at a, I guess, late stage at an b round startup called One Up Health.
So obviously for me, having been involved in doing the interoperability rules yeah, you, you would know it pretty well. In Inspire, one can imagine the incredible appeal of, uh, joining a group who's actually implementing . That really the back end types of infrastructure you need to make fire as transformative as Congress really designed the rules to be so very exciting.
Yeah. So one up focuses we're not. This isn't a commercial for one up. Yeah, but I'm more curious than anything. So one up focuses on the payer space mostly, or, well, we've started with a payer space and the payer to payer rules and obviously payer analytics. But ultimately it's really figuring out what do you need to make modern healthcare data extremely computable.
Right. So what is that? You know that, that sounds like marketing speak, right? So if you look at big data, right? The classic big data definition, nothing even to do with healthcare is large volumes of data, high velocity of data, and lots of variety in data, right? That's the classic big data definition. So if you look at that, you're then left with how do you actually implement that?
That's as I used to say in math class, non-trivial, right? Because whenever you have things where you need to be really fast, need to have really vol, high volume . And high variety. You're putting in things that are innately contradictory, right? I mean, they're just innately contradictory. What we have at One Up is a way in a totally scalable way in the cloud, and the cloud allows you to do these things.
Have combined a rich set of . Strategy so that we can take each element of data compute on it individually, which is really the heart of the magic. So rather than thinking classically of a database as a series of tables, and you iterate over the columns, you go down the roads, right? The classic, uh, relational database, what we've done is we index every single
A piece of data as a fire resource, and then you can go to town because the indexing allows you to put these together in any way, shape, manner, or form you want. So that can be things like, it could be classic sequel. It could be a modern columnar stores for very fast comparative computation. It can be no SQL rapidly indexing for an API for an app.
All of that is really leveraging the fact that in the cloud. With storage being essentially relatively inexpensive these days, you can put search. So finding the data in real time as your optimization, that's what really needs to power any modern big data thing. Whether it's providers reaching to patients, providers, analyzing data on their own payers, reaching out, you know.
Any business combination. So it's a, it's an enabler for that communication. Yes. But it's also an enabler. You could potentially, you know, one of the challenges is we have dirty data within AI models, . Oh yeah. Putting it down in that direction. Right. And I, I assume not the provider market yet, because the provider market is really required by the required by the EHRs.
e exact. Not to be exact. Uh,:Everybody's watching It is in the federal register. Yeah. So, so the. EHRs under the Cures Act. Right? So the CURES Act requirement was application programming interfaces, APIs without special effort, right? So not a vendor specific thing. The goal there was to let patients get their data on the patient's term, take it outta the EHR patients could do with it what they want.
So that obviously is, is, is a key component of, you know, I think modern computing app economy, competition, new ways of providing care, what empowering the patient. I mean it's the Cures Act, it's Right, yes. Empowering the patient right at the patient's choice and direction. Right. Right. If you're getting your data in a portal, you don't really have a choice.
Right. You, I mean, you and, and I can't do anything once I get it in my portal. Yeah, you either hand copy or something, but it's not, it's, and you know, that was what state of the art was, but you know, now it's, it's, it's different. But that is really one way the world, obviously the bulk fires for an entire population that the individual patient data.
Into an app, let's say, of the patient's choice is under the HIPAA right of access. There's also bulk fire, which will be transformative in a totally different way. The bulk fire API. So same US core data for interoperability. That is a batch mode transaction under HIPAA's treatment, payment operations, privacy provisions.
And that for the first time will actually allow providers or payers to really have a truly comparative view of how they're performing. Now, the providers may wanna do that. Let's say if you're a hospital network between sites, obviously the payers and providers may want to do that jointly on shared goals.
The payers may say, Hey, I'm trying to negotiate . A contract and network and I would like to see how you're performing. It's a very, that part of it, again, same data science revolution, but that uses a very modern, global sense of performance. Historically, for the last 20 years, we've had these explicitly narrow, heavily lobbied, largely honestly meaningless, though they were state of the art at the time.
Things that were narrow quality measures, very burdensome to report, but they were very isolated. Ultimately. Some of them were great, you know, blood pressure control, hemoglobin A one C, those types of things. But ultimately, they're not big data. They're, you know, teeny weeny data. And we can do, we can do better these days.
So it's how do you look at a population? And with the APIs, of course, all this can be done without burdening providers. So without having to have my notes as a doc, reflect my interpretation of what some email said about quality measures, I can use, you know, the institutional and my own skills to provide the best care possible.
Presumably that will show up in the overall product. So it's a very, very different view of doing data, and I think it is part of that modern way that fire will enable. So that's totally generic from the one up point of view. The point you mentioned that's critical is . How clean is the data? You know, everybody talks about AI and you know, we're here at HIMS and there'll probably be 50 talks,
On 80% of the work in AI is cleaning up the data, and 20% is on analyzing the data. Well, one up is really about the 80%. So instead of having all this heterogeneous data, very difficult to compute, we are. . Now we have a massively and subtly engineered pipeline to ingest heterogeneous data and convert it into fire so that when you, so the data is, oh, convert it into fire.
Yes. Convert it into fire. We'll take it if it's in fire, but we'll also convert, converted into fire. So then when you wanna do analytics on it, the entire data set is computationally way more uniform. And we can convert claims into fire. We can convert clinical into fire. And that's one of the subtle, under greatly underappreciated aspects of fire is rather than having a hodgepodge of data, things that almost by definition make it hard to do modern analytics.
By cleaning it up, up front and putting it into a uniform format, you have a much better chance of getting signal out, cleaning up the data for errors. . You know, some of the health equity things, you know, all those things require clean data. When are the payers required to comply with the 21st Century Cures?
Interoperability? Well, the, the payers are under CMS rules, so the payers aren't directly covered by, um, the ONC interoperability rules, the ONC interoperability rules. Or EMR vendors, providers, and networks. Okay. So unless a payer is acting, and of course increasingly payers are acting right in these pay provider.
Probably had a couple. So CMS came back and said, payer, so CMS said payers have to share same concept. We want patients to have their payer data too, as well as you know, their provider data. So the rules were time to sync up. Same concept, fire, bulk fire. So yeah, they're, it's easy to confuse 'cause they sort of are, I don't know, I'm not gonna get into any, uh, yeah, it's, it's different things.
Parent parentage things, but Right. Same. It's the, it's the same concept. And then as part of that, they're also some other pro-consumer efficiencies like payer to payer, you know, so for example, payer to payer is driven by the fact that if I'm changing, let's say. The dreaded aging into Medicare. Right? All of my prior, prior offs and all of that would would've been lost without a payer to payer interface.
When I change payers, right. They did all this stuff. I'm in the middle of my chemotherapy. I have to go through all of the approvals again. Right? So the payer to payer is part of that. For payers, we're seeing some real interest because they . Want to know what they're getting as patients enroll allows them to sort of get engaged much more rapidly.
And those are the rules. There's for payers, there's obviously, and for providers, there's a whole bigger issue, especially for payers of what are they actually doing, right? I mean, they're listed as insurers, but they don't really do insurance in the sense that a property casualty company does insurance.
They're really elaborate claims management and claims adjudication systems rather than insurance. And in that world, what data do you want? You know, right now, payers, . spending is, is, is, is fairly coarse, right? Prior auth, those types of things. Denials, you know, funny network design. There's some fairly primitive tools.
With modern computing, you can be vastly more targeted on how you use tools the same way as in the consumer world in the past, you know, the ad hit everybody. You know whether that person had a prayer in the world of buying whatever you were advertising. Now with modern analytics, all of the ads are targeted, right?
Same thing on the payer side. How do you target it? Historically, they haven't really had a great ability to do that. You know, and they're obviously population management program attempts. If you have your data in a clean computable format, you then have a much greater chance of having targeted intervention.
So the payer side and, and frankly, the same thing for providers, right? If you're a provider taking risk. Yeah, I'm thinking, I'm thinking of our clinically integrated network. We took risks. Yeah, yeah. And we had to, we had to measure performance against a certain set of metrics. And we also had to, uh, provide reporting out to back to the physicians.
And by the way, they were all on different EMRs, so that, that pipe would've been really, we tried to do it through the HIE not very effectively. And then the claims data came in it, I mean, it was, it's, I'm trying to describe this, it's . It really convoluted in a really difficult problem if you're trying to do a clinically integrated network across multiple EHRs and potentially multiple payers and payer models, that the data's coming to you and then your team starts to go to work once you just get this stuff dumped.
So I assume. Uh, you know, all these regs and fire, bulk fire was meant with that in mind to say, look, we've, I, I understand the consumer aspect of this. We want the consumer to have access to their data, but that, that clinically integrated network where people are communicating from one end to the other, providing the best care, I assume is the, is the use case that makes.
Really drives this whole thing. Yeah, precisely. So what you've described, and the way it's sort of been for anybody who's accepting risk is, you know, you have your claims on one side. They're not really integrated. Even skipping, you know, the delays in AR days and all that kind of stuff, they're not really integrated.
They don't tie together other than manually. So then your IT staff or whoever you've hired to do this, . I think you get a whole new set of vendors to do stuff, but ultimately you don't have the ability to have insights over time in what you're doing. And then because this whole process is so brittle, you know, people quote, fix it with reporting requirements, well that's obviously not a fix.
So if you think of this in a totally different way, which is let's ingest all the data over time. Have it and highly index things, then you can spend your time figuring out how do I improve care? You have analytics against uniform data sources and it also allows you to do this stuff over time, which is an important issue because the data process you outlined is so challenging.
It's been sort of a very much of a whack-A-mole, right? I mean, I think if you look at much of population health, it's been a bit of a whack-a-mole with modern AI techniques. You know, ltms transformers, not to get into all of the different buzzwords, but you can actually predict past to future, right? With these rich data models,
Enrich AI tools, you can actually pick the entire trajectory of your patients at great accuracy and then target the interventions you wanna do. Gives you a totally different way of looking at solving the problem. And then, you know, you write contracts and negotiate contracts to, to leverage that win. And obviously in a data world,
Whoever's fast and first is fastest. First and fastest and, and richest on the data. Yeah. Bobble that up is going to be the winner. Right? Right. I mean, it's like high speed trading. Yeah. I, we, we used to have this, I, I worked for a provider. Yeah. I remember saying that. And we would say essentially when we set across from the payers Yeah.
They had, they just had more data than we did so and so. Yeah. This gives you, um. On your population. Now, they obviously would have some other populations, but I don't wanna use upon fight fire with fire 'cause that would be too cheap, and lame. And I'm above that maybe. But yeah, it's, it's a world.
Especially when you, when so much of our network negotiation is really . Two large organizations going head to head. A little bit of a cage wrestling type of activity, right? It's not like a typical consumer shopping experience where there's lots of payers and lots of providers in many markets, both the payers and the providers of bulked up and there.
NC as national coordinator in:To quote Humphrey Bogart shocked actually to find out that the fire data standards were only for one patient at a time. It had never occurred to me that somebody would do a database and only retrieve one record at a time. That was like so stunningly non-performance, let call it was looking for a flight.
Lack lacked foresight. Well, it was what it was. So I'm talking with Ken Mandel, Boston Shoulder and he, he pointed this out to me and said, Hey, we should do something about it. And so we came up and I would've, if I had known that Bulk fire was going to bulk was gonna be the name that stick with it, maybe I would've.
Had something more harmonious, but bulk actually describes it. So, you know, I was told, maybe you should have called it population level data or something, but rate it's bulk fire now in the rule and the standards. But the concept is really pretty simple, which is we wanna look at the data on a population and not just one patient.
Now that's the, the exact same fire data elements, the US core data for interoperability. The EHR vendors already had to clean up and transform and to fire for the individual right. Of access. So minimal work, right? The heavy lifting already done. But so then you just have to have, you know, specification of what's the population?
The, the nature of getting that data is under HIPAA's treatment payment operation. So that's a batch mode. It's not real time like the individual Right. Of access. And it's negotiated. Right. The payer and provider, let's say if it's payer provider, could be, we heard, when I was at ONC, I heard a number of large providers couldn't even get their own data 'cause of contracts they'd signed.
Somewhat stunningly frankly, but whoever it is, if it's providers, obviously they have a right to their, to their data. This lets them look at it, put whatever tools they want. If it's a payer provider network, obviously that has to be agreed to, you know, today they're often like one off, one up, one off downloads of some SQL report that you can't really do much with when, which maybe is the intent of it.
That So you have a population that you've defined, that you have a right, a legal right to by, by contract, under covered entities, business associate provisions of hipaa. The power of it comes because as the US core data for interoperability, it's a uniform data set. So you can use off the shelf tools, you can have off the shelf reporting.
healthcare. Imagine we're in:Uniform way of analyzing population performance of data. Pretty stunning when you think about it, right? Yeah. So now with Bulk fire, we have that way, and I, I think it'll enable lots of, you know, analysis, ai, ml, all of these tools everybody talks about. I think they actually become practical. 'cause now you have data that is uniform
You know, if you go with somebody like us, you can have it be really homogenous, pre-processed and clean. But one way or the other, we can do actually all data, not just the US core data, but you're gonna have a start at a much richer, more reliable performant analytics. U-S-C-D-I. Yeah. Uh, US court data, you.
Do you keep saying it? It's, yes, it is full. The U-S-C-D-I. Where do you wanna see that go next? What data sets do you wanna see that tackle next? Yeah, so, so for the audience, USCD, I used to be called, the common data set is really the, the basics of describing a patient. I sometimes describe it as if it's changes shift.
Describing a patient that I'm taking care of to whoever's taking over for me, or vice versa. It's, you know, the basics of the history, the medication list, the problem list, the allergies. It has the notes. So it's more than that. It's essentially the family. It has, it has some provenance. Now the US version one, version two is.
You know, the current ONC is putting some other things into that, you know, from, you know, maybe more politically driven. But the US core data set, the initial one is really just that basic description of a patient with provenance on who did it. Interestingly, the Europeans are doing something almost identical.
Are they really? Yep. Yep. So in the European Union, there's a couple subtle differences, but almost identical, and it's that basic data that really defines the patient. That's interesting. Wow. So you're going to, this your chief strategy officer. What are you gonna do here at himss? What are you looking forward to seeing?
It's been a while since we've been together. Yeah. I'm not, yeah, it's uh, yeah. You know, HIMSS is always a interesting event to see, you know, people's view of the world, to hear what people are are doing, you know, it's. With the power of modern computing, I mean, the variety of what is being done by folks attending and exhibiting and speaking at hymns is, gets larger every year.
So it's a, it's a pretty, a pretty exciting type of thing, especially if you're a, a nerd. . Absolutely. Don, thanks for your time, bill. Thank you very much. Appreciate it. Much appreciated. . Another great interview. I want to thank everybody who spent time with us at the conferences. It is phenomenal that you shared your wisdom and your experience with the community, and it is greatly appreciated.
We also want to thank our channel sponsors who are investing in our mission to develop the next generation of health leaders, Gordian Dynamics, Quill Health, tau Site Nuance, Canon Medical and Current Health. Check them out at this week, health.com/today. Thanks for listening. That's all for now.