Claire Manneh of Datavant

Speaker A: Foreign.

Speaker B: Welcome back to Product and Health Tech, a community for health tech Product Leaders by Product Leaders I'm Chris Hoyd, a principal at AH Vynyl. Today I sat down with Claire Monet, head of provider research at datavant. We touch on several topics in the realm of health data connectivity and its impact on uh, healthcare research and outcomes. In our conversation we explore datavant's mission to connect the world's health data through their innovative protect, connect and deliver approach, including the power of tokenization in linking disparate health data sets while maintaining patient privacy and data security. Claire offers valuable insights into how datavant is working to break down data silos in healthcare, empowering researchers and healthcare providers with more comprehensive patient data. Let's dive in. Can you start by telling us a little bit about your journey into health tech and what led you to your current role as the head of provider research at datavant? Sure.

Speaker A: And first of all thank you so much for having me. It's a pleasure to be here Chris. I have always wanted to do something in healthcare. I was pre med, uh, was going through all the classes for med school and to go into med school and uh, realized after a couple of years from graduating that I wanted to do something that was impactful, uh, outside of actually delivering medicine and um, worked in bench research, clinical research, consulting. I spent several years working in patient safety and then got into tech, uh, working at ah, a uh startup that uh, did second opinions and then found my way to datavan. And what excited me the most about datavan is that this was a piece of technology that we had always thought about from years ago when I was on the clinical research side and it was about linking a patient's health history to follow their journey across the healthcare continuum. Um, and I've been uh, I've been at datavant for about four years now and every day excites me so much because I get to work with the very researchers, those who are physicians and those who are academic researchers who are making the impact in healthcare. And I'm excited because I get to be a small sliver of that as part of their work.

Speaker B: Okay, so Claire, for those who might not be familiar, could you give us a brief overview of datavance and its mission in healthcare?

Speaker A: Sure. Datavance mission is to connect the world's health data. We do that in three ways. We protect, connect and deliver. And they're right above me right now. Um, we protect the data because um, on the de identified side where I primarily focus my work in is to link datasets using A token. Uh, we're connecting the data sets by empowering researchers at life science organizations, at academic ah, med centers and patient advocacy groups to take their very private and precious data and either enhance it with another data set that's coming from the ecosystem like uh, an sdoh, a social determinants of health data set or, or a claims data set. And so they are connecting it and there are often times where we may need to deliver the tokens, not the data, but we may need to deliver the tokens, match them and then deliver it to whomever the data aggregator is.

Speaker B: That's pretty technical. It's like in the world of healthcare, it's fairly abstract. Can you connect that, you know, offering your product to population, health or health outcomes?

Speaker A: A really good example I can share. One that brings me back to the days when I did do clinical research is I'm working with researchers at the University of Wisconsin at Madison and they built a substance misuse data commons. This is a platform that brings together EHR data from the University of Wisconsin health system. There are local emergency medical services data, there are Department of Public Health, there's an all pairs claims database. They have social determinants of um, health data, mortality data. All of this data are combined in a platform that they created using the datavant token. And all the different agencies installed the datavant software behind their firewall. So we don't own any of the data, we don't house the data, we don't serve as a marketplace in that, that regard. But they've been able to link all of their data sets together and the honest data broker at the University of Wisconsin takes all those tokens together, links them, appends, uh, the, the, the necessary data that goes with it and has this complete history of patient data across the state and they're working to bring about other uh, data sets from other EHRs, uh, so that they can get a more complete picture of, of Wisconsin's population. And this is so exciting because now they're able to look at trends and follow what's happening at the EMS level. Are we seeing that there are an influx or any ah, trends toward certain types of substance misuse and looking at the health of these patients um, during treatment and after treatment as well.

Speaker B: What is exciting to you on datavance Roadmap? Sounds like you guys are increasingly integral to a lot of very complex stakeholder relationships. Is it, is it growth? Is it, is it new capabilities? What, you know, what's exciting to you?

Speaker A: It's everything that you just mentioned. I think um, it's exciting to see how many more hospitals, health systems, academic med centers that I work with. And then I also have colleagues who work on the registry side of the house. So anything related to a patient advocacy group, um, our government organizations too, all of them are coming together. Plus the data aggregators that we work with and life sciences. We're beginning to see a flywheel of all of those parties work together to better understand a patient's journey to improve health outcomes, to deliver better care, uh, to understand the differences between certain drugs, um, how they interact, interact with patients across different populations and regions. Health equity. Um, the other exciting thing that I'm beginning to see a lot of is the interest in unstructured data. Um, currently we work very closely on structured data. So de identifying uh, the EHR claims record, that's all structured data. But now especially with uh, an increase in the use of different AI technologies is taking the clinical notes, looking at imaging which we haven't tapped into yet, but hope to one day, but mainly on the clinical notes, um, is a space that I, I can't wait to, to see it come come to fruition for a lot of these studies.

Speaker B: Yeah, I was curious about that. So it sounds like you guys have, you know, there's no shortage of data, right? There might be the, the bottleneck might be turning that into, or your customers turning that into actionable insights. And so it sounds like you're hoping or expecting that AI might help with that. Do you guys do that internally or do you hope your clients will develop some wherewithal around that?

Speaker A: Yeah, so I primarily work with researchers at uh, academic med centers and health systems and they truly are the brains behind a lot of the AI, uh technology that they're working on. We also have AI capabilities within some of our tools. Our Match tool, uh, is built on a machine learning algorithm and it been trained on billions of public health data records out there. But other than that, um, most of the research is done by, by the researchers. Whether it's a life science company or uh, an academic med center. They're the ones who are, are really synthesizing the data after we've been able to help them link it together, understand it um, across different data sets. They, they really distill that and, and do their analyses and apply any models to them. The example I shared, the professor at University of Wisconsin Madison, he's been able to create machine learning models to better understand the data sets even on the structured side. So um, they all use, use their own own methods to understand it.

Speaker B: Before we jumped into this conversation you mentioned, uh, that you were really proud of the work that your team and the population health side of things. What you guys do. Um, outside of the Wisconsin example, is there anything else that you want to highlight within that realm that you feel good about?

Speaker A: Yes, I mean, I don't know where to start. There are so many different studies that we work with. Another ah, adjacent hospital, uh, to UW Madison is the Medical College of Wisconsin. They um, put together a registry for sickle cell disease and sickle cell disease is commonly found among black men in the US and there isn't a proper reg for this. They received a grant from the CDC among several other states to do this as well. They similarly have been able to link their EHR data to claims data, other SDOH data to better understand the population in Wisconsin, um, and then other types of studies that we work on. Believe it or not, there's still a lot of research around Covid. We're still trying to understand what's the long term effects of COVID and we're seeing a lot of uh, researchers still um, getting funding and getting refunded on their Covid studies. One of those studies, uh, that we had been a big part of is the National Covid COHORT collaborative. The NIH put together this uh, collaborative across almost 100 different hospitals in the US and they all linked their EHR data to this platform. And now that uh, we've done a lot of work around Covid, they've renamed it to be the National Clinical Cohort collaborative. It's still N3C um, but they are studying things beyond conditions and diseases, beyond Covid. And that's something really important for us because a lot of researchers can tap into that beyond the work of COVID If there's another type of um, disease that they want to look into, that's something that they can uh, reach out to that team to get a better understanding of their research.

Speaker B: Incredibly cool. Okay, awesome. So um, I think in a previous podcast conversation that you were on, you talked about global clinical trials as a potential future focus of the datavan team. Is that still a possibility? And if so, how are you thinking about the unique uh, challenges, ethical considerations that come with tokenizing and linking health data across different countries and healthcare systems.

Speaker A: It, it's very special because a lot of the hospitals uh, we work with have relationships with, with those at the NHS and, and then they have um, relationships with ministries of health and other countries, Australia, Japan, um, so, so we know that there are relationships there. We've invested a lot of time working on the HIPAA equivalent in Europe gdpr, uh, to ensure that we have um, software that is accessible and usable across different ecosystem partners that have health data or the hospitals themselves. Um, we're beginning to work on that now. Um, and then of course I should mention a lot of life science companies are based in, in, in Europe and they have offices uh, in Europe as well. So there is a lot of partnerships that we've already enabled and turned on in, in Europe today on the clinical trial side. But br, the academic side to the clinical uh, trials with the life sciences is still new and I'm still working through that. But hopefully in the next several months we'll begin to see um, those linkages actually happen. To answer your question about uh, what is required, what's needed to enable and activate those partnerships, um, we're working in de identifying the data and tokens that uh, and I should also mention what tokens are in more detail. A token is um, made up of different phi or pii, so names, dates of birth or addresses, zip codes and uh, those, those particular phi pii elements are masked into that 44 character hash string and that can be linked to other data sets. So we can work a lot across the alpha numeric numbers but we haven't been able that to a, ah, a different language, Arabic, Chinese, Japanese etc. Hindi and so um, so those we, we haven't worked in yet. But mostly um, the Roman Alphabet is um, primarily where, where we're working on this to be able to activate a lot of these partnerships. The considerations we need to be aware of are um, how our address is captured. Social Security is not uh, used outside of the U.S. but there could be other numbers, an NHS number. So those are the ways in which we're helping to develop uh, those tokens in other countries with our sort of

Speaker B: emphasis on product and health tech and the sort of big tent of product leadership and strategy and research. Uh, we do try to get a little bit tactical here. So I'm curious if you could just talk a little bit about what your maybe average day or average week looks like within datavank.

Speaker A: I would say a lot of the times it's um, talking to the customer, scoping new projects that they might have. Um, they may not know what we do in its entirety. Uh, a lot of the times they know that we link disparate health data sets together, but they may not know that we work with genomics partners. They may not know that um, we're doing the unstructured data for the Clinical notes. Um, so those are really exciting because I get to cross collaborate with my team engineers, product, um, those who work on our Privacy hub, which help us uh, certify the data as well and we obsess over that piece, our clinical trials team as well. I often bring the subject matter expert to a lot of these calls because they can speak to a lot of the work that we touch on outside of the provider and public sector space that I'm in. And then I have a lot of time that I spend talking to new providers and educating them on what we do. Um, these are mostly provider researchers and so a lot of times they either know about databand and they want to find a way to excel uh, their work, um, or they don't. Um, so I have to start from scratch too and it's not always easy explaining what we do. But um, anytime that you tell a researcher that we could link this data to that data, that gets them so thrilled and excited to learn more.

Speaker B: That's awesome. Okay, so how do you measure the success of your team's projects or initiatives? Are there like um, specific KPIs or OKRs that you check?

Speaker A: Yes. Um, so once we sign a contract after we've gone through the security reviews with, with the hospital or academic med center and they're ready to install the software behind their firewall, we want to make sure that they are successful in tokenizing, creating tokens on their patient population and they have access to a project dashboard. So if they're working on multiple projects, they could see uh, where the tokens have been created for a specific project. And the success metric is to know when everybody has been on board to link their data sets together. Um, so that example project I shared with you about Wisconsin, it's one data aggregator, a data coordinating center rather. And they are waiting to get links from all the other data sources. So making sure all those other data sources have tokenized and submitted their tokens to them is a successful metric to me. Um, I also want to know what is the status post connection of all these data sets? Have they been able to analyze the data and then are they publishing? Are they ah, having speaking engagements, doing poster sessions? I come from an academic world too, so I'm so invested in knowing what is the impact of our tools on the patient population. So I want to see the soup to nuts of um, the entire research to itself and, and bring it back to our team because I realize a lot of engineers and our executive assistants and, and others accounting team, they can work anywhere they want but they come to datavant because they know that the work that we do is, is very impactful in our community. So I want to be able to share back because we were able to activate these linkages and in this particular study this is what we learned out of the population in kind of this

Speaker B: era of escalating cyber threats, especially in the health care space. Um, how does data event fit into that? Uh, how do you guys think about that internally and how might you work with your partners to guard against that?

Speaker A: And this is a space that is quite constant and there's always going to be some type of cyber threat. I was on a call with a very large uh, health system a month ago and they had to pause their work with us because they were addressing a cyber threat that they had and it didn't impact our work but in the sense that uh, we, we weren't uh, involved in the cyber threat itself but they, they uh, they had another issue uh internally um, that didn't involve that, the tokenization process. So um, those are things that we are all aware of and alert and uh, we, we have alerts that we send out to let the team going on um, and then just explore is this impacting us in any way. Um, but thankfully uh, in, in any of the, the cyber threats that have gone on ah, at hospitals and, and other academic med centers, um, they haven't involved the uh, data vant tokenization. And I think this is one of the most important things that we obsess over. We obsess over data governance. How is data being shared in the hospital, um, who's accessing the the to? Um, I work very closely with the C suite at the hospitals because they also want to know who is using the software and how is this being managed. And um, the good news is that the uh, datavant software is something that uh, is, is encrypted, it's site specific. So when they are creating tokens they're not sharing anything that is going to breach any patient confidentiality. And so that, that's one of the reasons why um, we, we spend a lot of time and resources and in ensuring that the tokens are not going to release any phi or pii.

Speaker B: So let's assume you know, some years out that uh, datavant has been smoothly adopted by uh, some critical threshold of stakeholders. And so the benefits of your approach are starting to be starting to be felt in real ways. What do you think that world looks like? Are there any structural changes that would need to take place for that to happen from a Regulatory standpoint to a uh, you know, technology standpoint. Maybe it hinges on what AI becomes capable of. I'm just curious what that, what that dream state looks like.

Speaker A: I'm going to start from my, my little microcosm of this, this ecosystem that I live in and then take uh, it to, to the larger state. So within hospitals and academic med centers more and more I'm beginning to see multiple silos that are in hospitals exist at the much larger enterprise side of the hospital. So a cio, a uh, Chief Information Officer or Chief Data Officer will be able to create their own platform so their researchers can tap into that, use data in any way that it's available on a datavant token and empower the research that these researchers are working on. Today. It's siloed in such a way that everyone is on their own island and they're creating their own data sets. And oftentimes researchers at that level don't even know that the data is already there. And so if there's a way in which every hospital, every ANC can, can connect and have access to that data, that would be most useful. We're seeing on the government side the importance of bringing the EHR data sets, the patient advocacy data and any other types of real world data into some sort of platform that they may be creating that could be at the sub agency level or it could uh, potentially be one day at the NIH or the CDC at large. And I think that is going to help connect all those spokes to one place in a privacy preserving record linkage way. And all those other research studies could be benefiting from that data set instead of again all these multiple silos that we see today. Now zooming out even further with my counterparts on the ecosystem side, that's when life science organizations who, um, many of whom we work with, they'll be able to have partnerships at the agency level or at the hospital level and ensure that there's a way to get patients to a clinical trial and be able to identify where there are times and opportunities to get those patients to support them in a trial. And this is huge in oncology, huge in uh, adult congenital heart diseases and even in pediatric uh, cases as well in rare diseases. And I think this is going to help that flywheel move.

Speaker B: Now that makes me curious if you um, have any sort of personal beliefs about where headwinds might come from in terms of technological developments. Is it uh, AI and biotech creates more demand for clinical trials, what are you sort of excited about in terms of what you're seeing in the economy and ecosystem that might um, accelerate that vision.

Speaker A: I think the headwinds are mostly going to be on a case by case basis and having institutional review board IRB approvals consent for specific use cases. There are times where you can do health surveillance, population health research, um, without the need to consent the population. But when it becomes very specific at a clinical trial level there's the headwind of having to oftentimes go back to the patient and request them directly. And if you're working on a patient population, uh, from a health system that's trying to ask anywhere from tens of thousands to millions of patients um, if they would be open to that. So I think um, there needs to be a way to ask the patient from the get go uh, before establishing this type of platform to ensure that going forward they, they will be considered for, for any type of trial that, that can come up and, and help save their lives. And um, the other area of um, headwinds and challenges is um, making sure that everyone is aware of, of what this is. It's, it's still, it's still new to the patient population, to the consumer. The technology itself is not new. It's been around for, for, but trying to voice this over to researchers, to the patients like ourselves to understand its value is ah, something that, that's going to take some more time too. The researchers who we work with um, are quite pleased and happy to share this with our patient population. I think that is going to be one of the biggest benefits is ensuring that there's a feedback loop of the research going back to the patient community to understand this is the value that this type of data brings.

Speaker B: Listeners who might want to reach out to you or have additional questions about datavant, how can they get a hold of you?

Speaker A: If someone wants to get in touch with me they're welcome to email me directly claradatavant.com or they can message me on LinkedIn.

Speaker B: Thank you so much for joining us. You can also connect with us on LinkedIn, you can YouTube or on our website at uh productandhealthtech.com if you have ideas or suggestions on what you'd like to hear in a future episode or if you'd like to be a guest, please shoot us an email at Ah, info productsandhealthtech.com.