Ready for Regs: The Modern Data Stack for Healthcare is Here

In a discussion at the Da Vinci Interoperability Showcase at HIMSS24Joe Gagnon, CEO of 1upHealth, Erin Landau, Director of Product at Oscar Health, and Pieter De Leenheer, CTO at 1upHealth, discussed health data, interoperability, and the modern data stack. 

Erin explained how Oscar Health, a health tech insurance company, built its entire tech stack from the ground up, with the member at the center and interoperability front of mind. Oscar’s proprietary data platform is a mixture of homegrown tools and vendor partnerships, including 1upHealth. 

Oscar is leveraging its data platform to power its member experience in three key ways:

  • Using clinical data to power member experiences within Oscar’s mobile and web applications. 

  • Using data to build robust behavior change interventions to target populations, personalize communications, and measure outcomes. 

  • Leveraging data to hydrate customer service and internal technology to empower case managers and customer service reps to have better conversations with members.

Joe Gagnon, Chief Executive Officer of 1upHealth,  Erin Landau, Director of Product, Oscar Health and, Pieter De Leenheer, Chief Technology Officer, 1upHealth at HIMSS 2024.

Joe Gagnon: Well welcome. I’m Joe Gagnon, CEO of 1upHealth. I have a great panel with me. We’ll try and keep you entertained for the next 20 minutes talking about the modern data stack. We can make that interesting hopefully for everyone. I have with me Erin Landau from Oscar Health Director of Product. I have Pieter De Leenheer who’s our CTO at 1upHealth and we’re going to sort of dive right in. Maybe Erin can give a very quick update on what Oscar is and what you do there.

Erin Landau: Yeah, absolutely. Hi guys. My name’s Erin. I’m Director of Product at Oscar Health. At Oscar, we are a health tech insurance company. We have plans in the individual market as well as in Medicare Advantage. And I would say my predominant focus is in bringing all of our tech stack that we have built for the last 10 years to market via our +Oscar products. That’s my focus.

Joe Gagnon: Pieter?

Pieter De Leenheer: Yes, hello. I’m very excited to be here. I’m Pieter De Leenheer. I’m the CTO at 1upHealth and I’m responsible for the architecture and the strategy of our platform that makes it easy for healthcare organizations to get their data organized in the clouds and put governance on it and make it available for all kinds of applications downstream so they can use that data in a safe way, I would say.

Joe Gagnon: Alright, thank you for both. So I think we’re all walking around the show. A couple months after CMS published a new set of regulations sort of kicking the industry in the pants so to speak, to say that what we started a couple years ago with Patient Access was just sort of the appetizer to the big meal that we’re going to sort of dive into as an industry, which is we’re going to start to exchange data among all our parties, payer-to-payer, and we’re going to go from the provider to the payer and the patient’s going to have a real reason to access data, not just claims, because they’re going to be able to see a prior authorization approval or denial. They’re going to start to see hopefully over time, a more complete view of their patient information with longitudinal information. And so the 700 pages of rules that CMS published really are setting a foundation for what interoperability really will be.

It’s not file exchange, it’s actually the seamless movement of data between parties. It’s the building of maybe a longitudinal patient record that allows us to better understand who this person is that we’re serving. And it’s really the beginning of the foundation of the validation of all the concepts we have. We want to do value-based care, but we don’t have the data. We want to do better care management, we don’t have good data. We want to optimize on a member experience and we don’t have good data. And so over the next two to three years, we’re going to see that the foundation of data using FHIR as the infrastructure, using the cloud and the internet as the mechanism, is going to allow companies like Oscar actually to go and create what we would want to be a stellar member experience. So Erin, maybe tell us a little bit about how Oscar’s thinking about creating this stellar member experience.

Erin Landau: Yeah, absolutely. I would say the first one is really taking data, clinical data predominantly and powering our member experiences within our mobile and web applications. And it’s not just about giving members access to that data, it’s also about translating it and bringing it to them in a way that they understand and they can use that to actually make more informed decisions about their healthcare.

The second one is we are using data to really build these robust behavior change interventions to target populations, to personalize any communications, and to measure outcomes. And we’re using data to actually power that kind of end-to-end behavior change platform. And the third one is we leverage data to hydrate our sort of customer service and internal tools and technology to give to empower our case managers and our customer service representatives to have better conversations with members, to give them access to a member data, so that they can really improve the experience of the members when we are working with them on a case or helping them in a customer service setting.

Joe Gagnon: Oh, that sounds great. We all would like to be members of Oscar I think. Pieter, you have worked across data and tech for quite a bit of time, other industries as well. The more important data becomes, the more you have to think about data governance and data policies and data quality and all of these other concepts that people are talking about in the data world of a control plane and a data mesh and seems more and more complicated. But maybe you could talk for a couple minutes about, as we amp up, how available data is, how important it is to think about governance and management of that data.

Pieter De Leenheer: Yes, of course. Well, there are a couple of challenges to make sure that companies like Oscar can really consume robust data, robust data of high quality, complete data that represents inclusive population and do analytics on that and also try to drive behavior for the members. So there are all of these steps that need to be done there. 

We have seen transformative power in all of industries of data and today healthcare is sitting on probably the largest untapped data reservoir in the world. So contributing 35% of the compound annual growth of global data, that’s a lot of data. So the first challenge I think for companies trying to get it to gather that data because a lot of that data is scattered around, a lot of that data sits proprietary locked up into systems. And as companies are moving the needle towards moving to the clouds, they need to really understand very well of all this data where I’m going to put my energy, where I’m going to put all my investments in getting that data together and enriching it, because you cannot do it all of course.

So understanding really what those business needs are in terms of constructing a full picture of your patients and your populations that you want to understand is an ongoing process. So data management is not a one-step thing. You don’t get the data and you put it there. It’s a process. It’s a continuous evolution of understanding, okay, what data do we need to understand? How are we going to contribute to the patient’s records we’re building and how are the feedback loops downstream happening on how that data is being consumed? 

So that is really important and it’s actually CMS APIs are going to be great because it’s just another catalyst for starting and making that data move. And I think for companies like 1up, it’s also important for us to guide our customers through that digital transformation because healthcare is still early in the digital transformation.

I’ve spent a lot of time in financial services. Fifteen years ago they went through the same process and at the same time, the industry is also kind of volatile because of all of the merger and acquisition activity, which makes the digital transformation even harder, right? Because once you’re on a path towards like, yes, we have a plan now we have a program, you’re going through another merger acquisition where it brings a lot of new data sources and complications in place. So that’s what we want to figure out with 1up, providing that end-to-end control over that data. 

And, of course, last important thing yesterday also in the European Commission really implemented the AI law, which is going to definitely have an impact on how state legislation will happen in the United States in the long run. Just like we saw the impact of GDPR privacy regulations, which when you start combining more sources of data, a lot of that data is not going to be HIPAA regulated, it’s going to all be consent regulated. A lot of that data is going to be based on policies that are local. So how are you going to all bring that data together and make sure when you combine it, that you respect all those policies at the same time and automate that process with the right levels of transparency, explainability, and the right levels of access controls so that companies like Oscar can be successful in driving member experience, but doing it in the right way. So see, there’s a lot to be done for sure, but that’s kind of it in a nutshell, I think.

Joe Gagnon: There’s a lot there. Pieter, as you’re talking, it makes me think a little bit about that we now have sort of the permission to actually operate the industry the way we want. The federal government, which regulates not just how Medicare and Medicaid get executed, but it impacts how commercial businesses run as well because everyone has a combined book of business and we’re saying now that all this information can come together for the benefit of the patient and really the cost of operating. But, in that, there’s a lot of work that has to be done to think about as you bring these data sources together. It’s why FHIR matters so much, right? At least we have a standard that we can work with, a set of APIs that allow you to easily access that data. But how do we think about data quality? How does one feel when you start to share data? What are you doing to enrich that data throughout the process? Is that member matching? Is that completeness? What about the provenance, where the data came from? There’s a lot of elements of these new future architectures that make this member experience work because you want them to be confident. Any follow-up thoughts on that in terms of how these architecturally come together to make that all happen?

Pieter De Leenheer: Oh yeah. Data quality is one of my favorite topics and I would love to talk with Erin more about it, but data quality is, of course, partly dependent on how the data entry happens far upstream into your data value chain. So a lot of that depends on the interface where a clinician or a care manager enters the data into the system. That’s the moment where actually your data already gets contaminated, I would say, right? So you need to be able to have the right interfaces to enter the data. But even then, once the data is into the system, there are a lot of kind of mechanisms in place. There are alot of tools out there to build rules to discover data anomalies and all that stuff. But it goes back to the governance. It’s like what data quality problems are you going to prioritize? Is completeness more important than accuracy? In what context?

So over the last 10 years, companies have done a lot of metadata management, so they tag data with additional intelligence. So data stewards could actually be guided better to improve the data quality. Metadata management is a bit of a crisis now with, of course, language models that are supposed to automate that process, but still language models on their own also have their own data quality problems like our hallucinations and RAG, and all these other problems. So I think there’s definitely identifying the anomalies in the data, there’s a lot of technologies that can do that, prioritizing which data anomalies you’re going to resolve for what purpose, but then also make sure once the data goes into production, like a machine learning model or an application that you still can get feedback loops back, right? Because down in the application, the end user, like the member might continue to enrich that data and you want to capture that back into your system of record, in this case, the 1up platform. So you can continuously enrich that data as it comes back.

Joe Gagnon: That’s great, Erin, a lot there, right? Of course you nodded your head quite a few times. As you guys think about how you want to have this better member experience and think into the future and the role that the data is going to play in enabling that to happen, what kind of discussions are you having internally and what do you think about what’s next?

Erin Landau: Yeah, absolutely. I think there’s two big areas that get me really excited about Oscar’s future. And I think both of the areas are really centered around data access and interoperability as a whole. I think one of them is powering a greater portion of the healthcare landscape using our +Oscar products. So Oscar really was focused, our mission is refactoring healthcare, and we’ve been able to do that really well within our small sphere of influence within the individual market. But I think for us the next step is then taking what we’ve built internally and bringing that, 10-xing the impact, by bringing it to other payers, providers, and things like that. And I think payer-to-payer access and payer-to-provider access is a really key part of the upstream component of data flow into our products so that we can improve the experience for members, for not just Oscar members, but also for our client members as well.

And then I would say another big one that we’re really focused on is of course, sort of hear it all around the conference, is AI. And I think at Oscar, the strategy that we’re using is very much what we’re calling a continuous hack-a-thon approach, which is a build-test-learn cycle where every team from a technology standpoint at Oscar has an imperative to implement one to two new AI pieces of functionality throughout the entire course of the year. And I have also heard this as I’ve been going around the conference the past couple of days too, but I think really our initial focus is in back office automations and not necessarily bringing AI direct to interact with the patient or the member just yet, but we are definitely doing a lot of human in the loop-style AI components that we’re excited about, co-pilots with our Case Management teams and our Customer Success teams to bring to enrich their internal tools with AI, and bring improvements to things like let’s say claims tracing for example, or prior authorization is another big one. Or even synthesizing clinical data from EMRs and then surfacing them back to our case managers or the folks who are in our prior authorization teams as well. 

So I think those are really big areas that we’re excited about to see grow and I think are going to have a sort of big impact on the industry overall.

Joe Gagnon: It seems like you’ll be busy for quite a while. What’s really exciting is that for as many years as anyone’s ever come to HIMSS, we keep hearing about what’s going to happen and what’s exciting about working with you at Oscar is that it’s actually happening.

It’s no longer just this, we want this to happen in the future. You guys are actually executing and learning as you go. Any key takeaways at this moment? What have you guys learned recently about this moment that everyone here can sort of take away as they leave us?

Erin Landau: Yeah, absolutely. I would say the biggest learning for us is making sure that you have built empowered teams to be able to execute fairly independently. And I think what we have tried to do at Oscar is enable individual teams to be able to build, test, and learn and rapidly prototype on their own so that you’re failing quickly and you’re making smart experiment decisions and smart mistakes, and, of course, keeping security and privacy in mind, but really making sure that you’ve empowered individual teams to execute effectively and really experiment first in a way that I think we try to fail fast and fail quickly and try a bunch of different things rather than setting a super long-term roadmap and then moving really incrementally towards that.

Joe Gagnon: That’s great. I think we’re as an industry at a moment where it’d be good for all of us to be productively disruptive where we know that what we have today isn’t okay, broadly speaking. It’s not an indictment of any of the individuals, but the system itself is brittle. The system needs to become dynamic. It needs to be able to allow Oscar to innovate or any other health plan or health systems to innovate. It needs to become collaborative and not counterproductive to each other. The sharing of information isn’t going to only benefit us. We can’t use HIPAA as an excuse to improve the way the system operates. We need to understand that the decisions that were made really literally on our behalf by CMS and HL7 and others to give us now the platform. The clouds there and available for us. We don’t have to go and deploy that.

We all know how the internet works. We all know what we love about experiences. We need to bring that same level of inspiration that Oscar is bringing to the industry, but we all need to sort of push that. There is nothing really in our way. It’s just a bit of time now and some innovative thinking. And so thanks to you and what you guys are doing at Oscar. Pieter, thanks for your comments. We appreciate you guys spending a little bit of time with us and wish you the best for the rest of your show and your safe travels home. Thanks very much. 

Pieter De Leenheer: Thank you everybody. Thank you, Erin. 

Erin Landau: Thank you.

Share with your community

Sign up to get the latest insights and updates from 1upHealth