We welcomed Dr. Charlotte Hovet, NTT Data’s Medical Director of Global Healthcare Solutions to discuss big data in the medical world. She explores how it’s currently being used and the possibilities that lie just over the horizon.
This is part of our podcast series on big data thought leaders. Be sure to subscribe to our blog to get updates as soon as they’re published!
Transcript, lightly edited for clarity:
Andrew Brust: In the big data world, we’re fascinated by real life use cases and how data is actually being used. The medical industry is one that’s quickly realizing the potential that big data analytics has, and what it’s doing with data is in many cases incredible. Today we’re speaking with Dr. Charlotte Hovet, medical director of healthcare solutions at NTT Data Services, to hear from her how big data is changing the medical field. I’m Andrew Brust, and this is The Big Data Perspective. Charlotte, thank you so much for being with us here today.
Dr. Charlotte Hovet: Thank you, Andrew. It’s a pleasure to join you.
Andrew Brust: As I was saying in the virtual green room, it’s really nice to have a guest who’s focused on a particular vertical, a particular application and use case, and especially one as important as healthcare. Clearly all of us in the industry are eager that, instead of being in an echo chamber, we’re really building technologies that are getting applied for the good of an industry. If it’s for the good of humanity, so much the better.
With all that kind of flowery intro, my first question is, as much as big data needs to be sensitive to industrial needs, I suppose certain industries and certain disciplines need to be ready for big data technology itself. In your view, what changes does the healthcare system need to implement to make genuine, rigorous use of their big data?
Dr. Charlotte Hovet: All right. Boy, broad question, Andrew. Let’s break that down a little bit.
Andrew Brust: Perfect.
Dr. Charlotte Hovet: Let me just share that the perspective I’m bringing to this conversation is that of a primary care physician having practiced family medicine for 20 years, and then also being now in the world of clinical informatics for the past 10 years. Yeah, I’m bringing a primary care perspective to this topic of big data. It’s so broad. When we talk about big data, people think about all of these wearable sensors out there and the Internet of Things and how is that helping us, and so I think what we might do is kind of break it down a little bit and talk about that.
Well, what does big data mean from a clinical perspective?
From a clinical perspective, big data for me is really a collection of large and complex data sets which are difficult to process using our common database tools, such as the EHR. Obviously there’s just so much data out there. It’s very complex data. There’s all kinds of disparate data out there that needs to be pulled together. It needs to be analyzed and so forth.
There’s all this data out there that’s coming in from the wearable sensors and internets, and people are wondering, “Is that adding any value?” Well, I think it does in the short term. I think it adds value in the short term in terms of helping our patients, people and our community members do more self-monitoring and self-management of their healthcare. I really feel strongly that we need to be our own primary care providers. There’s no question that having all of that information out there, and having that data out there for people to use and manage is useful.
However, for most people after about six months, it starts to become noise. They start to lose interest in it. Then the question is, “Oh, so what do we do with that?” Then people say, “Oh, well, let’s, let’s forward it to the primary care physician. Let’s send the primary care physician all this information on, these wearable sensors and my Fitbit and how many steps I took a day,” and so forth. Yet, to the primary care physician, that’s just noise. Just data is noise. What has to happen as we move forward in healthcare, and as we’ll explain in this conversation, is that we have to turn that data into identifying outliers, finding things that are outside the normal pattern.
We have to take data and turn it into actionable insights. It’s really about taking data and yes, changing it into meaningful information, so that either the person at home can use that data to change their behaviors, to take action which helps promote health, or it’s information that the clinician can then use. The clinical team actually, because we’re really moving to a team-based care delivery model, where the healthcare team can use that information to take action in terms of caring for their patients.
Andrew Brust: Yeah, fair enough. Outside of healthcare specifically, when we talk about the Internet of Things, there’s often a breakdown between kind of the consumer view of what that is, which is often like smart devices in the home like service stats and so forth, and then there’s the whole industrial application where you’re talking about. You’re talking about things like antennas, cell towers that are out in the field, or equipment on a manufacturing floor. They all have sensors, and they all report data, but some of it is noisier and more kind of, I don’t know, pop data than it is directly useful in a discipline.
It sounds like you’re at least alluding to the possibility that, yeah, for medical devices, real medical devices, whether they be in the doctor’s office, the hospital, or sent home, that readings from there obviously are very germane readings. From Fitbits and similar devices, maybe less so. Then electronic health records, that’s probably not big data per se, but there’s still a high volume of it. If you can correlate these things, then maybe things start to become more useful and more navigable.
Dr. Charlotte Hovet: Exactly. We hear a lot about all about these wearable sensors and trackers and so forth. I think we’re now moving forward in that when we talk about biometric monitoring, in healthcare we’re actually starting to drill down where it’s becoming more useful. I know that at NTT Data, we’re doing some work with a client looking at biometric monitoring to help people with their athletic performance. We’re actually starting to analyze an athlete’s sweat in real time to determine their hydration and electrolyte levels. Then through this sweat analysis, you can actually improve their performance.
Andrew Brust: Is it really called sweat analysis?
Dr. Charlotte Hovet: It is called sweat analysis. That’s an example of using remote monitoring, using wearable sensors, all of this out there, to track an athlete’s performance. Those activities are going on with race car drivers where we’re using analytics to analyze all of their vital signs and g-forces and so forth. I think what we need to move to today in this conversation is using these tools to truly be proactive and prevent chronic disease because in the world of healthcare, it is chronic disease that we have to address. We have to begin managing chronic disease by taking a much more preventative approach.
I mentioned athletes and race car drivers. There’s so much that could be done in terms of pulling all kinds of big data together for detecting risk factors, risk conditions, for chronic disease before people have any symptoms. Having been in healthcare a long time, I can tell you that medicine has been a very reactive field. Now in the years I was in direct practice, we would be faced with a disease. We would treat it. If the pain got worse, we would alleviate it.
Today with value-based healthcare, we have got to be much more proactive and preventative. I can tell you that big data is key to this. It’s absolutely key to identifying risk factors for disease and using predictive analytics to really influence our decision-making as patients and as clinicians.
Andrew Brust: I’m thinking of an IoT analogy, right? We talk about in the world of IoT, Internet of Things. One of the applications that’s arguably had the most traction is this idea of preventive maintenance. It’s usually applied to mundane things, not like human bodies, but like elevators or heating, ventilation and air conditioning systems. Building predictive models based on all the sensor data could allow us to observe phenomena that tend to be precursors to breakdowns in things that wouldn’t necessarily be intuitive in that respect. It’s been really helpful there.
I hate to equate a machine with a human being, but it sounds like you’re saying there’s a similar approach and validity on the healthcare side where we can be much more granular in the data that we observe, and as such, we can be predictive and we can be proactive about treating disease rather than waiting for empirical symptoms to show up. We can look at precursors instead of just end results and resulting symptoms.
Dr. Charlotte Hovet: Absolutely. Absolutely. Let me restate that a little bit. The whole idea of value-based healthcare is to improve outcomes for both individuals and for a population as a whole. It’s to improve outcomes at a reduced cost. There is no question. I believe, like so many others, that the only way we can do this is through predictive analytics, and predictive analytics are basically based on big data. You have to have big data. You’ve got to have data from multiple sources, and those are often disparate sources.
In the future, yes, wearable sensors won’t be just Fitbits and other trackers that tell us how many steps we take and that we should get up and move and so forth. Those are important. Don’t get me wrong. I don’t want to minimize those, but I think moving into the future, we are going to be doing so much more automated monitoring of vital signs and blood sugars and all kinds of meaningful biometric measures.
Like seven out of 10 deaths every year in the United States are related to a chronic disease, and 45 percent of all Americans have at least one chronic disease. Think about that. About half of us have a chronic disease by the time we’re adults, and yet so many of those diseases could be preventable.
That’s why this predictive analytics is so important. So if we can give people information early on that will help engage them to change their behaviors, we can then make an impact on their overall health. As we move forward, we’re going to be doing. There’s the healthy population which, okay, you might do some monitoring of. Then you’ve got your rising risk population which are those people who maybe have one risk. Then you have your high risk population who often have two to more chronic diseases that are very complex. They will need the most monitoring.
They’re going to need to be monitored in their home on a daily basis, and that data all needs to feed back to their primary care team. I’ll use congestive heart failure as an example. Congestive heart failure is such a huge, huge disease issue in our country, and yet most of the care needs to occur in a patient’s home. That’s where we live. We live in our homes. That’s where patients should get their care. That’s where they should be monitored. That’s where the whole issues of remote monitoring and having all of that data feed back into the clinical settings, to the care team, that’s where telecare, telemedicine comes into play.
Okay, so that’s kind of the care delivery that goes on, but I want to get back to big data because with big data we’re going to take the information that’s in the EHR. We’re going to take the information that comes from all of these remote monitoring tools. We’re going to take the information, we’re going to be able to take our genetic data. We’re going to be able to take our microbiome data. We’re going to be able to take a lot of data and because of all the computational ability today and all of the algorithms that are being continually developed for predictive analytics, we are going to be able to apply big data and predictive analytics on a day-to-day basis which will truly, truly transform healthcare and outcomes.
Andrew Brust: Yeah, we have a term of art in the data world called data blending, where we’re talking about taking disparate data sets and not just doing the physical, tactical work of joining them, but actually being able to correlate them. It sounds like what you’re saying is that’s the horizon for healthcare analytics is being able to take these data sets that have been siloed, either because specializations within the medical field tend to have different systems and different databases or just because in practical terms that data was collected in different places. As you said, with chronic care, some of it’s going to be picked up by devices that are in the home. Obviously when it’s a clinical setting, you’re collecting data differently there. Of course, sensor data and office visit data are separate as well. Then you bring the consumer side into it, and it’s even more dispersed.
But what you’re saying is it’s the grand unification of all of these things that’s really going to have the breakthrough, and that even if some of the stuff seems a little superfluous, you’re saying no. The power that we have now is that we can cast a pretty wide net, and we can bring lots of stuff together. Therefore, we can probably correlate phenomena which we probably didn’t even have a hunch was related, and we might find relationships there. That could be breakthroughs.
What you don’t know is that I had questions specifically about chronic care. I had questions about the consumer devices and the biometric data. I had the question about what gets you kind of the most excited about what’s going on in the field right now. Inadvertently, you’ve covered all of that, which means I don’t have to ask those questions. My final question was around predictions, and even there, I think you’ve kind of covered it in that you’ve talked about how a lot of these data sets will come together and probably produce value from the whole that’s greater than the sum of the parts.
Let me follow up on that a little bit and ask you if you have other predictions for what may come down the pike in terms of practice and in terms of technology this year and into the more distant future a few years from now. How’s your crystal ball? What are you looking forward to? It can be stuff that you hope will happen rather than stuff you guarantee will happen.
Dr. Charlotte Hovet: Right. Again, one of the things I certainly hope will happen is a way that we really as a culture, a healthcare culture, that we really are open to sharing our data, and then obviously aggregating it and analyzing it and using it to forward patient-centered, information-driven healthcare. That takes a lot of different stakeholders. On this topic more specifically, one of the things that excites me is machine learning.
I’ve spent a lot of time recently in the world of imaging analytics. Again, we both know that big data is just one big term, and that it involves all kinds of data sets. Then we drill down to predictive analytics. Then how do we use all that data? I’ve learned a lot recently about imaging analytics. Imaging can be in the field of cardiology, it can be in the field of pathology, dermatology, but specifically, I’ve been working more in the field of radiology. There’s just billions of images that are stored today in archives. Let’s face it. Probably every one of us at some time has had an X-ray. Yet usually after the primary diagnosis, that image gets put away and stored.
The CT of your chest might have been done for one reason and that’s what the radiologist appropriately focuses on, but through sophisticated algorithms, you can look at those CT scans, look at all the pixels, and actually pick up other significant findings. They may not have been significant for that primary diagnosis for which the image was ordered, but significant in that it could be a risk for heart disease or a risk for this or a risk for that. You can use that data. You can then use that data to have some actual insights where providers can take this data then, and then as you said earlier, use it in a meaningful way on data that’s already been obtained. It’s already out there. It’s stored. Let’s use it.
I think that’s what we’re going to find. We’re going to start finding that there’s all these uses for information we have. There’s imaging data. Now we’ve got genetic data. How do we correlate that? Like if you’ve got a certain genomics profile, how does that correlate with the phenotype that’s recognized in your imaging studies? I find it really interesting how I think the future is putting all of these sources of data, whether it’s imaging data, genetic data, microbiome data, environmental data, and I think it’s going to be amazing as we aggregate and run algorithms using lots of different sources of data. We’re going to be able to identify risk factors and then from there, we’re going to be able to identify treatment models that really are precise for that particular patient, which of course comes back to precision medicine. We are going to be able to truly transform how we deliver care so it is very individualized. At some point, we’re going to be able to get rid of so many of these chronic diseases that today we aren’t effectively managing.
Andrew Brust: Your enthusiasm is definitely palpable. I want to contribute one observation and one more question, and I think that’ll finish this up. The observation is just that you’ve happened upon another analogy, not even for big data, but for older business intelligence technologies. It’s always been about this idea that we have kind of business systems of record that are collecting transactional data for whatever the business may be. It could be sales. It could be office visits. It could be particular procedures that were administered and so forth. It’s one thing to collect it. It’s something else to then really, really analyze it and learn things from it. That’s been true in the mainstream data world. It definitely sounds like it’s a huge factor in the medical world as well.
You’ve mentioned a couple of times just the fact and the situation, the circumstance on the ground, which is that we have a lot of this data collected. It’s sitting there. There’s a lot of benefit to be had from correlating it and aggregating it. I don’t know if you meant to reference this or if it’s just something that triggered in my brain, but there’s ethics questions about how that data is handled, who has access to it, at what level of granularity or aggregation it’s going to be okay to share and publish it, and so forth.
This by the way also exists in the data analytics mainstream. It’s not specific on the medical side, but it’s probably more even acute there because obviously this kind of data is very personal to people, and willingness to share it or trepidation around sharing it could be pretty high. Could you tell us what you’re seeing in the field or what you expect to transpire in the field around focusing on the data ethics, because it’s a pretty big question overall? My observation is that the industry overall hasn’t done much with it, but it wouldn’t actually be too hard if we only focused on it. What’s going on on the medical side?
Dr. Charlotte Hovet: I think to answer that question, clearly ethics is always an issue, and making sure that people’s identified information is private and secure. It’s so important. That’s why we have HIPAA. There’s, I think, the word, what is it, anonymized, so the information is de-identified. We do need to make sure that that’s the first thing. I would say with this ethics question, the first thing is to protect people’s privacy and the security of their healthcare data information. That’s important.
Then, yes, I think that like anything we need to ask people for permission to use their anonymized data. I think most people want it used, especially when it can help through especially, we haven’t talked too much about what machine learning means, but how machines obviously can analyze all this data in ways that human beings simply cannot. As brilliant as our brains may be, we cannot analyze these volumes and volumes of data, and so that’s the benefit today of machine learning which is a form of artificial intelligence.
I think my experience has been that, if you explain to people and you can to the best of our human ability protect and secure that information, they want it to be used to better outcomes, to just improve healthcare, because I think we all know that there’s always room for improvement in healthcare, in the way we diagnose and in the way we treat patients. I think most people want it to be more personalized, precise care. I guess that’s my long-winded answer to the fact that I think if you inform people of how the information is going to be used, what it’s going to be used for, I think that we can manage this in a very ethical manner.
Andrew Brust: Okay. Excellent. We’ve covered an awful lot. Like I said, it’s been nice to get away from just the technology side of it and really, really see a use case that obviously everybody can identify with because health is important to everyone. It’s also interesting that we saw a lot of analogies between that particular application and some of the more generic questions that we’ve been discussing in this series overall. I want to thank you very, very much for joining us. It was edifying, truly.
Dr. Charlotte Hovet: Thank you, Andrew.
Andrew Brust: I’ll wish you a great rest of 2017.
Dr. Charlotte Hovet: Great. Thank you, Andrew. It was a pleasure. I look forward to conversing again.