Healthcare Rethink - Episode 16
We have now entered a universe of “determinants of health” ranging from medical debt to housing to medication access to transportation… literally all things impact health that happen outside a doctor’s office. On this episode of Healthcare Rethink, Claire Manneh, Head of Provider Research at Datavant shares how the intuitive approach of tokenizing data is transforming how the healthcare ecosystem is reaching patients and improving lives!
Episode Transcript
Brian Urban (00:22):
Yes, this is the Healthcare Rethink podcast. I am your host Brian Urban, and today our episode was spurred by an interesting question: Can tokenizing data actually improve individual health? Well, I am no expert at all on this, but I have someone who is. Joining us from Datavant is the head of provider research, Claire Manneh. Claire, welcome to our little show.
Claire Manneh (00:51):
Thanks for having me, Brian.
Brian Urban (00:53):
This is so exciting because we've talked about doing this episode for a little bit here, and we're finally here, we're finally doing it. So with each of our episodes, Claire, we like to get to know our guests and share a little bit more about your background to our audience to familiarize who you are and what you do. So you're at Datavant, a little shop there in San Francisco, California. You have a huge impact across the world though. But how did you get here? Take us back to what led you to Datavant and what led you really to now being the head of provider research there?
Claire Manneh (01:30):
Sure. I always had an interest in health care. I initially wanted to go into medicine, I was pre-med. And with the exception of a year and a half stint working for a hedge fund, I've exclusively worked in health care, and always wanted to do stuff behind the scenes. One of my first jobs actually out of college was at UCLA, and I worked on a couple of opioid studies. These were NIH studies where I worked on the intake, doing consent form for the participants on the studies. And this was all the data that we had. It was whatever we collected from that. Whenever they came in for the visits, we'd record everything, compile all results and share it. So one of those studies was actually to look at the effectiveness of Suboxone, which is the drug that's used now to support people who have a likeness for opioids.
So anyway, that was one of my first jobs and that was all the data we had. Fast-forward to all the other work that I did, I did more clinical research, I worked in consulting and patient safety. And when I came across Datavant, which was a newer company, it's been around for about five years, I thought, "Wow, if only we had this technology, the ability to link disparate health care datasets together, we could have learned so much about that research, that trial that I worked on many moons ago at UCLA." So I was so fascinated by that. And then in between all of that work from when I was at UCLA and working on all these other health care studies and projects, we could have also used Datavant as well. So here I am, and I am so lucky and fortunate to work with a number of different provider researchers at academic medical centers and health systems at collaboratives, and they get so excited that this technology is available so that they can learn more about their patients. So that's what I get to do today.
Brian Urban (03:49):
That is awesome. And I have to brag about you as well because UC Berkeley academic background and then of course onto Dartmouth for your master's in public health, I just think that's such an interesting blend because you've done the research, now you're doing really these partnerships you're leading and then also the innovation side of your work at Datavant as well. So I have my little hat here for you because we share some lone pine green colors.
Claire Manneh (04:23):
Go green.
Brian Urban (04:23):
Let me to do that for a second if it's on there. I love the Dartmouth Institute. We always got to be proud of that, leading some great research there themselves. Speaking of that, Claire, I wanted to get into a little bit more about what tokenizing data is, and then we'll connect that into the research that you're leading now. But I understand tokenization to really be, you're replacing a sensitive piece of information with a surrogate value or a descriptor, and that's very different than encryption, which is like scrambling data. But can you level set for our audience what tokenizing data means? Because that touches your entire solution suite and feeling all of your research as well.
Claire Manneh (05:08):
Sure. That's actually you're on the right track too. Tokens are essentially a key for a patient. It takes the PHI or PII elements from a patient record that could exist in an electronic health record, in a registry, in a social determinants of health data set. And it pulls those originating PHI, scrambles it into a 44-character hash string, and creates this new identifier or identification key to be able to link it to other data sets. So you won't know that this is the data from Dartmouth Hitchcock Medical Center for Brian Urban, it'll be a 44-character hash string. It's a two-step process. It's hashing an encryption, as you had mentioned, and using an algorithm to develop that to create it.
I love one of our customers actually was able to illuminate what this looks like to someone like my parents who are not in health care and technology, and one of the analogies that they used is making a smoothie. So you take whole fruit and you take fruit, you take ice, you could put your protein powder in there, blend it all up, and then you end up with this scrambled fruit smoothie, but you can't go back to the originating whole fruit again. So that's essentially what we do at Datavant, and I think it's one of the coolest things because all the tokens that are created are site specific for each project, for each site. So all of it is done behind the firewall that our provider researchers do. So they do it on their own, they create the key themselves, data doesn't get sent to Datavant. So they do that on their own.
Brian Urban (07:03):
That is a very good analogy, I would say, that could be universally accepted. And that's just so helpful, I think, for everyone that's listening to this to get a better understanding of not only what Datavant does, but the process in which you work through to be able to secure sensitive information but also enable it so you're finding the individuals you need to help or providing the next step in a very important clinical research study, et cetera. So, good walkthrough. Thank you, Claire.
I want to talk a little bit more about Datavant and then into some of your work. So I think a big announcement obviously, acquisition of Swellbox, patient accessing their own information. That's huge. What's really interesting aside from that is the Switchboard platform. I think this is so unique, and there's a really good visual on your website that shows just how the tokenization process works and the interconnectedness of a lot of unstructured data that can help start to make sense. Maybe even profiles or social health profiles is where my head goes, what your data could be used in the SDOH world. So can you tell us a little bit about Switchboard platform? And I want to get into really how your work is touching health equity in a second. That's what I'm really excited to talk about.
Claire Manneh (08:23):
I'm excited to talk about that as well. The beauty about Switchboard is that it's end to end. How do you start? How do you finish? I think one of the important things that we spend a lot of time working on at Datavant is compliance. We obsess over HIPAA requirements, we have the privacy hub piece of this whole Switchboard, and it's, how do we make sure that we are not at risk or compromising any of the identities of these patients? These are people who are studying, technically not, we the researchers who are studying. And there is an algorithm, a statistical algorithm that these statisticians, data scientists work on to make sure that when you take one data set and link it to another data set, that there's not going to be any compromised data that leaks out that is for a patient.
So expert determination is one of those two forms of the Health and Human Services Administration's ways of protecting patient privacy. Safe harbor is the first, which is deducting 18 different patient identifiers of data, but that can't be linked and it's only that data set that you get to use. Expert determination, part of the privacy compliance portion of the Switchboard, is what we like to obsess over. So I think we start with connecting the data, making sure that what we have is something that can be connected whatever we start with, the data sets that the providers start with, that that data can be used to create a token, and then going through the process of compliance and being able to deliver that together.
Brian Urban (10:21):
I'm glad you mentioned that because it's, I think, the world we're in. Advancing data as a service really and different platforms, the whole tech world, you think about fast, fun, impactful innovations, but underneath that, the fundamental piece that has to be there no matter what is compliance. And if you're not securing information in a legal way, in a compliant regulatory way, then everything else is going to kind of crumble down. Building your house on sand versus on a rock kind of analogy. You have to do it right, and doing it with integrity is extremely important. So thank you for being able to elaborate on that.
All right, Claire, let's get into the really impactful stuff, speaking of. I want to understand how your work is connected to progressing health equity overall, US, globally. We talked earlier about your relationships with over 60 academic medical centers including federally qualified health centers. I mean, that's a huge network of very valuable, high talented researchers that are all focused on helping improve human health and helping bridge gaps. So I want to understand how your work connects into that.
Claire Manneh (11:42):
You're talking about PCORnet, so thanks for bringing that up and leading me into that. The National Patient-Centered Clinical Research Network is a collaboration of over 60 academic medical centers and health systems across the US. They've come together because they've found that by partnering together and linking their electronic health record data together, that they can look at comparative effectiveness research. As one of those offshoots, some of the research that's come of that is, a group from Cornell and from Children's Hospital of Philadelphia wanted to better understand the effects of COVID to enhance recovery, especially for long COVID. So they created a subnetwork of 40 different PCORnet sites, and this is an NIH funded grant that they're working on, and they want to understand how to prevent this, how to treat this. This is all happening live right now. As you know, everyone is still trying to figure and understand how can we better work through this pandemic. And that's what Weill Cornell Medicine and Children's Hospital of Philadelphia are doing in partnering with this network through the PCORnet collaborative as well.
Brian Urban (13:08):
That's amazing. I think it's very difficult for at least the US landscape to separate their ultra-competitive business models and healthcare models and put that aside and really focus on how can we collaborate to better improve the healthcare economy, how people get access to care, how are we identifying people that have needs so we can reduce their use of ER services, and work more with family medicine and other preventative medicine too. So that's just so unique. I love the example too of CHIP, of Children's Hospital in Philadelphia. They had such an interesting background from where they came, which was having the worst health outcomes, the worst inequities for children and for adults in America dating back into the '50s, '60s, and probably before that. And now where they are, they're leading a lot of medicine and how you treat families and how you get children healthier and better services and better access to services. So great, great example of that.
You had mentioned to me earlier along those lines, your work with University of Wisconsin-Madison, there's a substance abuse project. You even talked about your work dating back to UCLA. So bring me up this view here, how this project is going, and what you're doing right now.
Claire Manneh (14:33):
Yeah. Thanks. I think that's one of the most exciting things, is to be able to come full circle to see where you started and where things are working towards. So being able to have that background, working on substance abuse when I first started at UCLA and then partnering with the University of Wisconsin. Dr. Majid Afshar is our critical care physician, a pulmonologist by training as well. He's building the first of its kind informatics platform to bring both state level and county level organizations. They're using the Datavant software, linking it to the University of Wisconsin electronic health record data and other claims mortality data. Other stakeholders are participating in this to better understand substance misuse prevention, how to treatment map across the state and improve care delivery.
What's really exciting is that Dr. Afshar has this informatics background. He's part of AMIA, the American Medical Informatics Association. He's developing these algorithms and applications that are centered around their substance misuse data commons to inform those patient level and population level interventions. And I think that that technology piece is super cool to see how is that going to really help expand the scope of what they're doing to prevent substance misuse in the state. There is a focus on opioids, but it's also alcohol, it's also methamphetamines, it's benzodiazepines as well. So it runs the whole gamut as well. I think it's also really important to highlight that this is bringing the Department of Public Health, it's the county, the Department of Corrections, the state PDMP, the claims data sources, all of them are collaborating together because they see the value in being able to bring all of these data, link these data together in a privacy preserving way, and better understand how we can help these people in our community.
Brian Urban (16:54):
I love this project because one, it's focused on good work. And the other thing is I think it's an accelerant, and maybe I can just ask you this boldly, to potentially building a mental health and substance misuse infrastructure in America. I mean, you think about if I fell right now, broke my ankle, an ambulance would be here soon. Or if I was very rural, helicopter, I would get flown to a hospital, treated immediately, build the insurance company. That's a business model. But mental health, if I had something right now in terms of sharing depression with you, good luck, I guess, right? I mean there's hotlines, of course, which is great, but there's no infrastructure in place. So do you think a lot of these projects and this work with these data will lead toward building an actual infrastructure of mental health and substance misuse in America within maybe a decade plus? The hope is there, but I mean, do you think these projects will lead toward that?
Claire Manneh (18:01):
Yes. And I use that example, the Wisconsin substance misuse data commons as that example, because the ability of using that machine learning technology will get an advance of what may come, look at the trends in advance of any type of situation from becoming exacerbated in the state. And I think that that technology and being able to use those algorithms that he's working on will help treat in advance and be able to find the patient and support them and give them any type of services that they can use and take advantage of before things get worse, before they end up being incarcerated, before they end up dying. And I think ultimately that's what a physician, a nurse, anyone who's working in the clinical field wants, to make sure that they help people get better.
Brian Urban (18:59):
Yeah, absolutely. I mean, I couldn't agree more with that comment in terms of maybe how this technological leap through machine learning will help maybe just go on top of the very old sick care model we have in healthcare in US and just be a newer, better component than maybe having to build an infrastructure from the ground up. It's just something new that can get puzzle fit and then hopefully adopted across the US. But you got me excited. You got me jazzed up about that because we need that type of work to actually be fulfilled and actually come to fruition.
So Claire, you said something, let's stay in the Midwest, the Upper Midwest, the frozen tundra for a minute. We talked about Wisconsin project there. University of Michigan, you mentioned earlier to me about an ophthalmology project that's going on. And I love what you said. I won't take the words from you. Tell me a little bit about your perspective in ophthalmology connected to overall human health and then what this project is with University of Michigan.
Claire Manneh (20:02):
Sure. And here's my ode to Michigan too, the hand, where are they? Where's Ann Arbor in the hand? So University of Michigan has an ophthalmologist, Dr. Josh Stein, who has created this network of his own built from the ground up, a very grassroots level type of need that he found to better understand ocular diseases in the US. He wanted to see what patterns of eye care look like just outside of Michigan itself, Ann Arbor too. There are so many different changes that can happen and he thought, what else can we learn outside of the state? So he's connected with a number of ophthalmology programs across the US at other academic medical centers, and they want to help better understand what does the quality of life look like for patients with ocular disorders.
So the database that he has built is de-identified using the Datavant tool, the tools that we have, to link all of these EHR data across the different academic medical centers. The researchers can access this for their own research use as well. And everyone gets use of the Datavant software to tokenize their patient population and profile. What I really wanted to say about this too is, it's one of those senses, our vision that we do take for granted. It's one of the most beautiful senses that we have, that we can see and visualize. And I think that we need to have some more attention on this too because what if we were not to have our vision too? I think it would be really hard and difficult to have life. And these are a lot of the patients he works with, especially ones who go through glaucoma. So he really wants to help improve that eye care in the US.
Brian Urban (22:06):
That's amazing because you think about population health relative to ocular health. I think about the homeless population in America, and you think about how we don't have a benchmark on how good their vision is and how that's impacting their quality of life, homeless and then through their journey of maybe getting out of homelessness. I'm sure that project will lead to a lot of really good things. But I guess, is the population being studied in just large segments, age-based or is it geographical-based? Is it socioeconomic status? I'm curious a little bit more about the population, if you're able to allude to that.
Claire Manneh (22:50):
Sure. These are insured patients. So it is the patients who are going to Penn Med, to University of Michigan Medical School, to Stanford, to UCSF. We don't know a lot about the homeless population, unfortunately. And I think that that is an area that we can certainly look into and tap into.
One of the coolest data sets that is out there or types of data that's out there is social determinants of health data, and this will really get at the crux of health equity and understanding what are patient's propensities to buy alcohol from a grocery store and actual groceries versus other processed foods. What is their education level like? Do they rent? Do they own? Do they have a consistent place to stay in a home? All of this information is really useful and we can also have a better understanding of these patients. Have they come out of homelessness or are they going into it too? We've seen a lot of people, especially during COVID, to have a complete change in their socioeconomic status. So I think being able to study that is something that a lot of these social determinants of health data sets can be attributed to the existing electronic health record data that they're working with today.
Brian Urban (24:23):
That's really helpful. And I think that's where I was hoping this would go. It's good you have to start somewhere to understand the connection of ocular health in terms of disparities there, and then connecting into the SDOH insights. That's huge because that gives a very good sense of what people need. I am the biggest critic of any organization in a healthcare ecosystem saying risk, member risk, or risk stratification. Oh, my gosh. It's what people need. It's just human needs. And how do you help other humans? We do enough of underwriting risk in the insurance world. What are the people needing? I think there's a lot of great health equity leaders coming into insurance organizations now. Sachin Jain leading Scan Health, he's amazing leader. Bechara over there at Kaiser Permanente. They're just amazing people that are talking about human needs. And I've never heard the word risk come out of their mouths because they don't see it like that. They see about helping people. So I like how you touched on that specifically to connect all those big dots there. And that's just so exciting.
So Claire, Datavant, you work with over 15,000 clinics, 120 health plans, and then specialty pharmaceutical and genetic labs as well. It's so much work and you've touched a lot of it. Can you tell me maybe what's your proudest moment or what's fulfilled you the most? It's probably tough to pick one, but just gut response here. What comes to mind of the most impactful work that you've touched so far? And then I want to look into the future a little bit too.
Claire Manneh (26:08):
Yeah. I think every researcher comes to us and they want to have a better understanding of their patients, and I really appreciate that when they do share that with us. To pick one project, I think it's a little tough there. We did... Gosh, we had a researcher, touching on this subject too, a researcher from the Medical College of Wisconsin last year. She wanted to better understand the homeless population in Chicago. So she took the services, some of the social service programs that were available there in Chicago, and who accessed them. And she was able to get an understanding too through county, the Cook County medical records and other health system medical records on these patient populations and get an understanding of what services they were using and taking advantage of and looking at those trends. She hasn't published yet, but as soon as she does, we'll begin to learn a lot about that. And I thought that that was something that was really special that she particularly worked on.
I think anything that touches children too is also really important. We don't know a lot about how COVID affects children on a greater scale too. And I think all of those are in progress. I also can say, and I have so many too, that's why I'm going to just throw a smattering of them, the University of Texas at Austin has created an app to help support people in their community, in vulnerable communities, so those who are at risk for housing, those who are at risk for using substance misuse. And they created an app and partnered with a social determinants of health organization to help them locate where's the nearest helpline or hospital or clinic or service center that they could take advantage of to improve their work skills or to get a food bank access to a food bank. So I think that particular project is also super special as well.
Brian Urban (28:39):
Yeah, I agree with you on so many fronts of pediatric care, children health research. That's where my heart goes for a lot of efforts inside FinThrive and outside my research world too. And this is excellent, Claire. I just love learning about this. And you actually work with Children's Health of Phoenix pretty closely, and we can use them as an example or anyone else in the pediatric world. But I guess, how is Datavant helping providers see their patients differently? And we could talk a bit about this maybe on the research end a little bit if you want, but how is Datavant helping change the way providers see their patients?
Claire Manneh (29:25):
Yeah, let's take a step back too. As consumers, Brian, we have our health data at the hospital. There's the EHR there, but we could also have gone to a lab to get our blood drawn or do other types of samples there. And then we may have wearables too. Our data is everywhere. And if you're like me, you've moved several times, and so you've gone to multiple hospitals and clinics and urgent care perhaps too. So your data is everywhere.
If you are a provider or an academic researcher and you only have that health data that you either collect through a survey for your patients or from the electronic health record, that's essentially all you know about them. What we help researchers do is to be able to link all these disparate health data sets together because they're so siloed. Health care is one of those things that we're just that little sliver trying to help improve health care in this way as a tool. And I think that is going to help illuminate the way that a researcher can see their patient history and look at the different journeys that this patient has gone through across their years of life.
Brian Urban (30:50):
That is really, really insightful to hear. It's the journey that a provider can see along an individual's progress through health or regression in health or changes in life, adverse life events, losing loved ones, losing job, things like that, moving, all kinds of different events. That is really helpful. It's going to change the way I think probably.
I was just talking to Dr. David Nash of Thomas Jefferson University, who's at the College of Medicine there. He was describing how the education construct needs to change for those in pre-med and those in medical school to incorporate more of a medical humanities touch to it, a social determinants of health, and having that as the bedrock of education, how you interact with patients and how you consume data to determine what the conversation is like with a patient. That's next generation. That's where health care needs to go. Thank you for sharing that. That's very, very insightful, Claire.
I want to take a leap into the future a little bit here. So Datavant touches the whole healthcare ecosystem now in terms of your research work, providers, health plans, pharmaceutical, and genetic labs that we were talking about as well, and all these really interesting projects at different levels of public health, community, and county that you're giving examples of. Two, three years from now, what do you think Datavant's most impactful work will be touching?
Claire Manneh (32:30):
I would venture a guess that it would be something to do with the global healthcare world. So anything that touches patients across the world, clinical trials probably, I would venture to guess that we would tokenize all of clinical trials worldwide to better understand what is a patient's journey across different countries and being able to look at their healthcare within their health system versus what it looks like in the US, in Japan, in the UK. I think if we're able to expand that scope, that will be most useful.
The reason why I bring that up, Brian, is because I lost an aunt to cervical cancer, and it was so hard to try to find a trial for her to join. And by the time we learned about that, it was too late. There are so many trials out there that can support people going through not just cancer, but all types of rare genetic diseases. And if there's a way that we can help link all of those patients together so that they can get the treatment or the possibility of getting some sort of treatment, I think that would be the wave of our future too. So I would do clinical trials on a global level.
Brian Urban (33:59):
Thank you for sharing that, Claire. I think that's a very passion generated place where Datavant is coming from. And having leaders like yourself, being able to have these amazing research projects in your hands, great partnerships and innovations coming up, we need more Claire Mannehs in our world. I'm glad you're doing the work you're doing, and thank you for sharing that. And man, just an exciting horizon for Datavant. But I just really enjoyed talking about the whole landscape, everything you touched. Just a wonderful conversation. So thank you for joining our little show, Claire.
Claire Manneh (34:40):
Oh, this is such a pleasure. Thanks for having me, Brian.
Brian Urban (34:43):
For more insights and excerpts from our conversation, please visit finthrive.com.
