Ìý
Episode Transcript:
Dr. Frank Maddux:ÌýThrough data analytics, data engineering, research, collaborative partnerships, and publications, Â鶹Éçmadou Medical Care's Clinical Advanced Analytics Team uses real world data collected as part of routine dialysis treatments to generate insights that can be used to support clinical activities, product related efforts, or data science related projects. Allowing clinicians to use artificial intelligence algorithms as part of their routine clinical practice.
On this episode of Dialogues, Dr. Len Usvyat, the Global Medical Office's Head of Clinical Advanced Analytics, talks about how real-world data is used to improve the care of people living with kidney disease.
Welcome Len.
Dr. Len Usvyat:ÌýThank you. Thanks so much for having me here.
Dr. Frank Maddux:ÌýI thought we might start today just a little bit describing, Why do you think kidney disease is such a big data kind of enterprise? How is it that that big data has become such an important part of the world we live in?
Dr. Len Usvyat: In nephrology, and particularly patients who are on dialysis in that segment of healthcare, we're very fortunate because we do collect so much data because we see patients so frequently. I can't think of other segment of healthcare where you can see patients for four or five, six, seven, ten, 20 years, even, sometimes routinely, not just once a year and not just once every six months.
But we actually see patients three times a week or maybe even daily, and we're collecting data on them, and I think that makes it very unique. I think dialysis, kidney disease is very unique in that aspect that we have this audience of patients that we're able to collect their data. And I can't think of other disease where we're collecting so much information about our patients on a routine basis for many years in a row.
Dr. Frank Maddux:ÌýHow are the various ways in which that data is used both to effect a treatment, but also to understand individuals or populations?
Dr. Len Usvyat:ÌýBecause we're collecting so much data, there's a number of things we can do with it. I would probably divide certainly our team’s activities into these two main categories. One of them is real-world evidence generation. So, the data that we're collecting through these various data systems that we have, not just in the U.S. but globally, we're using that data to generate insights.
Those insights can be something that we can share with our medical community, with patients, through publications. We can share it with a broader set of clinicians. So that's the real-world evidence piece. It's learning what we know from the data. The other way we can use this data is through data science and turning the data that we have into insights that can actually be used in a routine practice at a at a dialysis clinic, at a nephrology office where we can actually generate insights, real time for our providers to be able to make decisions based on what we're seeing in patients data using statistics, mathematics and a variety of other advanced tools that we can utilize.
Dr. Frank Maddux:ÌýDo you have any project over the past ten years that you think has been sort of your favorite example of how advanced analytics has actually made it to the bedside?
Dr. Len Usvyat:ÌýIt's always hard to pick one because, you know, it's always hard to pick one child as a as a favorite one. But I think I would say I think the hospitalization predictive modeling work that we have done has started many years ago. And I think it started in what I would now look at as a pretty naive and fairly basic way. And It has evolved over the years through our learnings, not just by improving the techniques that we use from a data science or machine learning perspective. Yes. And we're also using more data than we did six, seven years ago. But I actually would argue how we implement these things in real clinical practice. We've improved and learned so much on how we're using these models in real practice. And I think that's the piece that I should really stress is it's not just about using this data and data science, it's actually about applying them in real clinical practice learning and iterating and making things better as we move along. I think no model is perfect, but you can improve as time goes on.
Dr. Frank Maddux:ÌýClinical staff will understand the typical types of data we have, like blood pressures and heart rates and you know, weights and things like that. But over the years you have thought up some pretty novel types of data sources, not just public data sources. But I'm thinking back on the concept of the acoustic fingerprint and the fact that even though we started in one direction with that, it's probably found other uses. You want to just describe some of the novel sources of data that you use?
Dr. Len Usvyat:ÌýMost people would say: oh, great, we have so much patient data. Why do we need something else? Why do we need another data source? And I would argue that there are some very innovative ways of collecting data and gathering data that you don't even think about it. You know, we're surrounded by data as we're sitting here or, you know, in any other setting. And so, for example, we back in the day when we’re predicting no shows for our patients, we would leverage things like weather data that's coming out from NOAA. We used, as you mentioned, acoustic data. We tried to use this sound data that might be coming out of our dialysis clinics to understand whether there is something in the dialysis clinic that may be changing just purely based on the sound, not listening to people's conversations, but really just the data that's coming out, the numeric data that's coming out, de-identified numeric data that can provide us some insights.
Text notes are such a great source of insight that back in the day we didn't really have technology to be able to process tens of thousands of text notes and now we can just put it through some NLP algorithm, and that algorithm can give us some indication of whether there's something in that text note that is more indicative of a patient, for example, going to the hospital. There's many, many different data sources, you know, fitness trackers, all kinds of ways that we can use the data a little bit more intelligently. And I know, you know, Dr. Peter Kotanko always uses this visual which I certainly love, where we have a lot of data on our patients, but in reality, it's only about 12 hours during the course of seven days that we actually have information on our patients. So there's a lot of unknown stuff to us still about what may be happening to the patient that can be used to improve their outcomes.
Dr. Frank Maddux:ÌýHow important do you think it's going to be in the future that we gather data from the environment somebody lives in as a way to understand what's happening in their life and what some of their health needs are going to be.
Dr. Len Usvyat:ÌýIt's a wonderful data source. And I think if we can overcome some of the, of course, privacy and other concerns that exist out there, but do it in a very smart way where you none of us are interested in, again, I think when we're talking about acoustic fingerprinting, none of us are interested in hearing the conversations that people have. But it's really more about converting that into some sort of a numeric digital wave that you can then utilize to give us insight about the patients or, as you mentioned, about the patient's home. And I think for home patients, even particularly for home patients, but even for in-center patients. But for home patients, that's a huge question of, you know, how do patients live?
Is there some sort of a high variability in loudness? For example, at night, maybe the patients don't sleep well because they're living next to the highway and they're hearing all this traffic noise. And we would never really think about this unless we have all that data somehow collected in a de-identified way. So there are lots of things we can look at when it comes to patient homes. You know, their temperature that the ambient temperature that they live in, maybe the sound that's coming out of their environment all kinds of other things that we can look at. So, I think it's very important.
Dr. Frank Maddux:ÌýYou and I've been interested recently in talking about the environmental impact that our therapies provide and whether it's on energy utilization or water, gas, whatever it may be. What are some of the novel uses as a company that we might think of analytics influencing how we provide our treatments, interact with our patients, manage our colleagues, and some of the novel things that you and your team are doing.
Dr. Len Usvyat:ÌýIt's a combination of things. One is data. Thinking of innovative ways of collecting data and I think you brought up a great example of think about what may be happening in the patient's home that we don't necessarily collect right now, that we should be able to collect not by doing another questionnaire., because I think we all I mean, yes, we can do questionnaires, but I think, you know, there's only that much willpower.
People have to be filling out questionnaires but being smarter about how we collect that data using devices like microphones or other, you know, thermostats, I mean, all kinds of technologies that we may have at home that are allowing us to collect that data from patients home. In a dialysis clinic, it's the same thing. We know from the environmental data that in the U.S., we use air conditioning a lot more than I think we probably do in most other parts of the world.
I think how does that affect the delivery of dialysis treatment, knowing that our ambient temperature in the dialysis clinic may be a lot lower here than it is, for example, somewhere in Europe. This is one of the many, many things that we could be we could be looking at when we think of more innovative ways of delivering care in our clinics.
Dr. Frank Maddux:ÌýYour team and other teams have done a lot to contribute to our representation at ASN. Can you just speak a little bit about what some of the work has been done and other things that you all have focused on.
Dr. Len Usvyat:ÌýAs you can imagine, ASN is a big time for for our team. We all get excited. We all get excited about ASN given it's the probably the largest compendium of nephrologists worldwide in one meeting. Our team has over 20 different abstracts that have been accepted, and I think there's also a couple oral presentations, and it's a variety of topics. It's hard to describe each one of them, but maybe I'll give you a couple of big themes. There's certainly work on COVID. As you can imagine, we've done a lot of work, Linda Ficociello certainly, and our team and others have done a lot of work. In looking at what is the antibody response to various COVID 19 vaccines in our population.
I think we all know what the response is in the general population. But understanding dialysis population response is something that we've been wanting to do, and we've been able to do. There's a few abstracts that we have. We have accepted now at ASN dealing with antibody response level. We've done a lot of work in looking at phosphate binders and how do we how do our patients respond to different phosphate binders? How does quality of life of patients change if they are or they're not on a certain phosphate binder? We've done some work in on the COVID side. We've done some very interesting work in Europe that Anke Winter and her team has done at looking at what is the mortality rate in the patients who had COVID, but also what was the mortality rate in the patients who were never diagnosed with COVID?
And we've actually found that there were also patients who never had COVID. Their mortality rate was also a bit more elevated than what it used to be in the past. Obviously we all know there's a number of factors that could play a role. Maybe it's undiagnosed disease, but it could also be just general healthcare capacity issues that many patients have experienced during COVID times. We've done some work on for this initiative called Inspire. There's a few abstracts that are accepted for the initiative.
Dr. Frank Maddux:ÌýWhat is the INSPIRE Initiative?
Dr. Len Usvyat:ÌýThe Inspire initiative is a consortium that we started a couple of years ago, includes a few well-known academics, mostly in Europe, as well as Â鶹Éçmadou. And the way they Inspire Initiative works is the group together comes up with ideas of what we want to analyze in our data.
We have lots and lots of data, and we want to make sure that others, academics, nephrologists, other folks, are able to leverage those data resources in a smart way. And so, the way we've set up the Inspire initiative is that while the questions may be coming from the Inspire consortium, we typically do all the analysis in-house. It makes it much easier. We started about a year ago, year and a half ago, and, so far, we have one manuscript pretty much finished. Then the second one already in the works. So it's a very exciting project.
Dr Frank Maddux:ÌýWhat are they about?
Dr. Len Usvyat:ÌýThe first question we wanted to understand was related to GI bleed and GI bleed events. And so what we tried to do is we tried to say, can we build a predictive model using our various data science and machine learning algorithms? Can we build a predictive model to identify which patients will have GI Bleed episodes?
And we build such a model, it has hundreds of variables in the model. We are also hoping to pilot that model at some point in real clinical practice, as I mentioned. I'm not only excited about publishing something, but I'm also excited about bringing those things to the patients and to the nephrology community. We've also identified some interesting things like Vitamin D level and its association with GI Bleed episodes, which was a little bit surprising to us. So we have a couple of abstracts from that area.
Dr. Frank Maddux:ÌýI have a couple of projects I want to specifically get you to comment on. One is the Apollo database and just give us a little sense of what we're trying to do with the Apollo database, what it is, and and why you think this will be important.
Dr. Len Usvyat:ÌýWe have lots of patient data, but having a lot of patient data globally doesn't necessarily mean that we had a way to easily access this data on a global level and to be able to easily consolidate our findings on a global level. We had some ways of doing it, but there's no straightforward way because we all know that every country has its own privacy regulations. So, as being in the U.S., I can’t just log in and be able to analyze data that's coming out of Europe because, of course, Europe has GDPR regulations, and they're pretty strict about what we can do with the data. And of course, vise versa. Because of HIPAA, we can't just have somebody in Europe analyzing our patient data. Given that we have so many countries where we provide service, it would be really fascinating to be able to understand what is happening to our patients. If we combine these data sources and we can do these things a little bit more transparently and faster and more efficiently than saying, we have to do 43 analyzes from 43 countries because everything has to be done individually.
The idea behind the Apollo database was to say: well, can we create an anonymous dataset that's coming out from most of, or all of our dialysis clinics that we have globally? And it becomes an anonymous data set. So it's no longer governed by things like HIPAA or by GDPR because it no longer applies because it is not considered patient data anymore once it's anonymous. And so The Apollo database is a way to do so. We're in the process of almost completed, thankfully, of creating a database that would have nearly 400 variables per patient. Hundreds of thousands of patients would be included in this dataset. And I think 43 countries would be in this dataset and the dataset could be analyzed very easily. So I would be analyzing the data set and I could easily stratify by country, by region where the data is coming from. And I could, for example, see if the COVID predictors, for more severe COVID response are different from country to country.
Dr. Frank Maddux:ÌýAnd it's longitudinal data isn’t it? it's not just a slice of time.
Dr. Len Usvyat:ÌýThat's right. It's a It's a three-year dataset. And I think We can extend it beyond, of course, three years. But Initially it's a three-year dataset. So, it will have, from an analytic standpoint, It is so important to have longitudinal data because just giving me one row per patient is not really that helpful, I mean it gives you some information, but what's really important is being able to look at patients over time and this dataset would be longitudinal.
Yes. It's not going to have dates per se, because that's not something that I think either HIPAA or GDPR would allow but wouldn't really need dates. For most of the things we do, we don't really care what city the patient came from because that's not really essential to understanding clinical questions. But it will have most of the information that we would need. So I'm very excited when it happens.
Dr. Frank Maddux:ÌýHow important do you think it is that we pair this clinical phenotypic information sort of what's happening to a patient that we observe or we measure? And then, as it's happening, to genotypic information and how an individual is made and built, it seems to me there's a growing body of interest in combining both longitudinal phenotypic data with genotypic data.
Dr. Len Usvyat:ÌýGenetic data is another data source. It's a It's a giant data source as we all know. It has a lot of information in it. But I think it's extremely powerful. Combining the data that we're collecting in our various EMRs that we have globally, together with data that's coming out of genotypic data I think would be very, very powerful. Back in the day, we were interested in combining claims data together with clinical data. Now, I'm very excited about being able to add the genetic data to this data set as well.
Dr. Frank Maddux:ÌýHow does the MONDO database fit into this larger picture of looking at granular treatment related data? What differentiates that from Apollo?
Dr. Len Usvyat:ÌýThe MONDO dataset is a little bit different. The Apollo database is purely an FMC Â鶹Éçmadou creation. It's something we've been wanting to do internally because I think it's important for us to be able to do analytics and research more efficiently on a global data set.
The Mondo dataset is a little bit different. It does include some of the Â鶹Éçmadou data. It doesn't include all, but it includes a subset of Â鶹Éçmadou data. But what's critical there is that we have a number of external dialysis providers, non-Â鶹Éçmadou dialysis providers, academics, who are also part of the Mondo Initiative. And that's very important to us, of course, because we want to be able to leverage all the expertise that exists out there. The data set has some similarity to Apollo. In some ways, it's a little bit smaller, but it's wider in that it includes data from some of the other providers outside of Â鶹Éçmadou Medical Care.
Dr. Frank Maddux:ÌýYour team comes from all over the world and, probably much like other teams, but certainly representative of our company, it's highly diverse and a lot of different ways. And I'm just curious how you describe how everybody works together, having come from different backgrounds, different languages, different systems of both healthcare, and cultures.
Dr. Len Usvyat:ÌýI think diversity helps. Having different opinions, having people come from different countries is actually very, very helpful. My challenge as my team has become global, has been how to maintain some of the efficiency that may exist because of the more local things that people are able to do versus how to also leverage the fact that we're a more global team, which means that we have expertise all over the world in various matters that some people might not have had. One of the big things I've been trying to do is, is trying to figure out that right balance between being able to be stay efficient. But on the other hand, also being able to leverage skill sets that we now have globally or expertise that we now have globally that we never really had in the past. Fortunately, now our team does have access to data globally while we're staying compliant with all the various regulations that may exist in a more local level. There's been a lot of learnings that we have had over the last couple of years as this team has been able to globalize.
Dr. Frank Maddux:ÌýWhat do you think the most important topics are that we need to address that we're not we're not addressing at this point?
Dr. Len Usvyat:ÌýI think one of the key challenges that I often think about is really how to integrate our advanced analytics into truly the clinical workflow. And by that, I don't just mean sending out, you know, have building another dashboard or sending out another spreadsheet, but it's really how do we create solutions that are used within our dialysis clinics, EMR solutions, or other solutions that we have where those data science driven techniques are actually integrated into the workflow?
Where, to the nurse or the clinician or the physician, it’s transparent. They don't have to go to another system to find out what if somebody has a risk score, they know that, you know, this is the risk score for hospitalization. And maybe there are some recommendations that are being provided and the physician thinks like it's all very well integrated. To me, iPhones have done such a wonderful job in being able to be very smart about how they all of that is integrated into one device and hopefully we're getting to that point at some point where our EMRs will be able to do the same thing for our patients.
Dr. Frank Maddux:ÌýYou've had, over the past ten years, a fair amount of work that you've done with our value-based care segment and so forth. As you know, we recently combined Â鶹Éçmadou Health Partners, InterWell, and Cricket. And I'm curious whether you see that as a huge opportunity to even go further into understanding what are the things that are most important in patient care?
Dr. Len Usvyat:ÌýAs you can imagine, of course, historically, our team has been mostly focused on dialysis patients who are in our dialysis clinics. Now, I think with InterWell and Cricket, obviously the focus now is not just on those patients, but also to we want to be able to focus on CKD patients who are not yet on dialysis. And I know that Cricket of course has done some really wonderful work on engagement and other tools that they've been able to develop over the years. So, I'm very excited to see how those things integrate. Now, I realize on CKD patients, we may not have the same wealth of data, but we just have to be smarter about how we use the data that we have.
It kind of goes back to your original question. I'm I think it's very beneficial to be able to combine these different efforts and improve how we deliver care, not just for dialysis patients and FKC clinics, but also for CKD patients and dialysis patients where we may be taking risk on, but they're not in our clinics as well. So all those different patients.
Dr. Frank Maddux:ÌýI'd like to congratulate you and all the people throughout the company that have contributed to getting back to ASN this year. And both the work that we've done as a company and individuals within that company to present ourselves there.
Dr. Len Usvyat:ÌýThank you. And I know they've worked very hard and I know they're very excited, some of them at least, who are able to go to ASN. There's a component missing, I think, in when some of these meetings are purely virtual. So I do think there's a benefit in kind of maintaining this kind of hybrid structure.
Dr. Frank Maddux:ÌýI'm here today with Dr. Len Usvyat, who is head of our Clinical Advanced Analytics team in the Global Medical Office. And Len, thanks for being here.
Dr. Len Usvyat:ÌýThank you so much. Thanks for having me.
Ìý