(Ed. Note: The “Innovation Fact of the Week” appears as a regular feature in each edition of ITIF’s weekly email newsletter. Sign up today.)
People have been working on digital transformation of health care for decades, but the COVID pandemic has added urgency to the challenge. This is creating a window of opportunity to reinvent how the health care system works—for example, the United States is on track to surpass more than 1 billion virtual office visits by the end of the year, even though only about a quarter of health-care organizations offered virtual visits before the pandemic. But of course, we still need to deal with important issues like privacy, security, and usability of all this new technology. ITIF VP Daniel Castro joins regular Innovation Files host Jackie Whisman to discuss these issues with Pat Combes, worldwide technical leader for health care and life sciences at Amazon Web Services.
- Nigel Cory and Phillip Stevens, “Building a Global Framework for Digital Health Services in the Era of COVID-19” (ITIF, May 2020).
- Daniel Castro, “Improving Health Care: Why a Dose of IT May Be Just What the Doctor Ordered” (ITIF, October 2007).
Jackie Whisman: Welcome to Innovation Files. I’m Jackie Whisman, Vice President for Outreach at the Information Technology and Innovation Foundation. We’re a DC-based think tank that focuses on technology policy.
Daniel Castro: And I’m Daniel Castro. I’m Vice President of ITIF, and I direct our Center for Data Innovation. This podcast is about the kinds of issues we cover at ITIF, from the broad economics of innovation to specific policy and regulatory questions about new technologies.
Jackie Whisman: In this episode, we’re focusing on the use of data in healthcare and the life sciences. I really love this topic, and it’s one you’ve been working on for a long time, Daniel, so I’m really excited that you are here to cohost with me.
Daniel Castro: Thanks, Jackie. I’m happy to be here. And you’re right, I spend a lot of time thinking about the future of data-driven medicine, because using technologies like artificial intelligence and machine learning can help us expedite drug development, improve patient outcomes, and lower healthcare costs.
Jackie Whisman: And we’re recording this, of course, in the middle of a global pandemic. The COVID-19 crisis has exposed very quickly the importance of precision medicine and innovative treatments. We at ITIF have been talking about this for a long time, but it seems like a lot of people are just suddenly educating themselves on not only the life sciences pipeline, but also might now be more willing to make use of different healthcare apps and telemedicine options in this more isolated world.
Daniel Castro: Absolutely. Many people have been working on digital transformation of healthcare for decades, but the pandemic has really accelerated this change. This is creating a window of opportunity to reinvent how the healthcare system works. For example, the United States is on track to surpass more than one billion virtual office visits this year, even though before the pandemic only a quarter of healthcare organizations even offered these virtual visits. But of course we still have really important questions to deal with, like privacy, security and usability.
Jackie Whisman: Well, that’s why we have our guest here. He’s going to give us some perspective on all of this, and I’ll introduce him. Pat Combes is the Worldwide Technical Leader for Healthcare and Life Sciences at Amazon Web Services. Before AWS, he spent time at Fermi National Lab, Cray, Merck, Microsoft, and EMC, so he has a ton of experience engaging with users and partners in the medical technology industry. Thanks for being here, Pat.
Pat Combes: Thank you.
Jackie Whisman: What are some of the most impactful ways we are seeing data being used in healthcare and life sciences today?
Pat Combes: Well, I would actually break this question into those two areas. The impact is a little bit different in healthcare and life sciences. On the healthcare side, we see data being used to really identify and drive tailored treatments for individuals. This is especially significant in the pandemic that we’re going through right now, in that we’re able to quickly identify vulnerable populations or underserved populations very quickly and easily and accelerate their treatment based on the kind of population group that they’re in.
And on the life sciences side, we’re seeing, actually, something quite magical. If you look at the usual time that it takes to develop a vaccine or even measure the times between when a condition, a pathogen has been identified and a vaccine enters a clinical trial, what we’ve seen in the past five months, six months has been unprecedented. That is a data-driven exercise on the life sciences. Now we are at a point where we’ve got about five or six potential vaccines entering clinical trial right now. This is really a magical transformation of the industry that has been very data-driven.
Daniel Castro: You bring up data, and one of the key changes we’re seeing right now in the sector is that so much more health data is moving to the cloud. How does this improve security and interoperability for different health organizations?
Pat Combes: Well, the first thing that it really does is that if you look at the cloud, we always think of it as being far away and sort of ephemeral. But if you look at the types of breaches that you see in healthcare, for the most part, these are physical breaches, and these are the types of breaches that can’t really occur on the cloud. Instead, when data is placed on the cloud, it comes in a number of very restrictive and extensive controls are applied to that data as it comes in, versus the sort of reactive controls that are applied to data when it’s sitting on-prem. For example, as data is brought to the cloud, it’s always encrypted in transit and it’s encrypted at rest, just sort of a first order application, as opposed to what you’d see applied on-premise, where we apply controls and security measures reactively to data as we acquire it on-prem.
Jackie Whisman: What do you see as the biggest barriers to greater use of data in healthcare? I mean, it seems like there are many.
Pat Combes: The biggest barriers that we see in greater use of healthcare data are really the healthcare data itself. There was a talk that I attended a number of years ago describing a particular but very widely utilized health record system that indicated that even accounting for other conditions, amputations, accidents and things like that, the average height as described by this EHR health record system, the average height of an individual over their adult lifetime varied by nearly a foot. And what it points to is really an issue with the quality of the data that’s gathered in healthcare. It’s inconsistent. It’s often locked away in a lot of different silos that the kind of measures, inputs, integrity of that data and so on and accuracy of that data just really isn’t there and becomes its own barrier to greater use of that data. If we can’t rely on it, if it’s not accurate enough, then we can’t utilize it for a lot of the cool stuff that we intend to do with it.
Jackie Whisman: And would you agree, I think better use of data and analytics, if we can figure out that piece, could really help bend the cost curve?
Pat Combes: Absolutely. Even in the simplest sense, if we look over the industry and we can develop better ways to provide tools and things to reduce the number of redundant tests or things that can shorten the time to diagnosis or treatment by utilizing reliable data that can be found in a patient’s health record, then that will necessarily reduce the cost that we all encounter in the health system.
Daniel Castro: One of the things that I think is exciting in this space is that we see health data apps that are emerging. They’re become much more common and they empower patients to access and control their health data. What types of opportunities do you see on the horizon in this space?
Pat Combes: This is one where I would actually refer back to the data quality and accuracy problem that I described earlier. By giving patients the ability to take control of their own healthcare data, they can become the agent of that change in quality or that improvement in quality of that data. And what I mean by that is that if you look at how you manage your finances right now, and you access this data through mobile applications and things like that, if you saw just unbelievable differences in the way your finances were reported, incredible discrepancies in there or anything like that, data that was inaccurate, you would go and be the agent of that change. You would go and initiate some type of reconciliation or correction on that data.
By giving patients the ability to take control of their own healthcare data, they can become the agent of that change in quality or that improvement in quality of that data.
And a similar thing is at work here. You can’t really imagine that the patient is going to be able to always interpret the result of tests or anything like that, that are presented to them by these applications so much, but you can imagine that they become the agent of change that is driving better quality in that data. By making it accessible to them, it really highlights the either discrepancies or gaps that exist in that data and motivates the patient to really fill them in.
Daniel Castro: One of the things I think about in this space is that we saw this big change occur with these apps in the financial services sector, where banks had all the data, and now we have all these personal financial tools that let you manage your money. That’s what we’re seeing start to emerge in the health system. Can you talk about the role of cloud in making that possible? Because it seems like this API access to data is what’s enabling this whole new third party market of apps and services that can give patients the tools so that they can actually manage their healthcare information.
Pat Combes: Yeah. The API-accessible data is just the door that’s presented on it. Really, the fundamental change that the cloud represents in that case is the elimination of silos. If you look at patient data as it’s represented right now, if you try to reconstruct your own health record based on encounter data that is spread out over a number of different silos and a number of different geographies, potentially you grew up in one place but you live in another right now, and so your health record is sort of spread all the way across these institutions in a physical sense. The cloud provides a common platform or a common home for a lot of this data and breaks down those existing silos as it is and really is the core change that enables a lot of access to this data.
Jackie Whisman: COVID-19 has forced many healthcare providers to rely more heavily on digital infrastructure. To what extent do you think this will accelerate digital transformation in the industry?
Pat Combes: I think what we’re all seeing right now and all the way across in many different sectors, not just healthcare, is the compression of a lot of change into a very short amount of time. Maybe we’re seeing 10 years of change happen in only 10 months, and what that has really pushed forward is the use of a lot of different digital, not just transformations, but applications and features and value that these transformations provide. When we get back to normal and when this current crisis ends and we get back to a sort of a steady state operation someday, hopefully soon, what we’ll find is that a lot of the value that has been provided by these digital transformations are tools that we’ve all come to rely on, and they can’t be just replicated in paper process or anything like that.
These are extra access to information or quick access to powerful analytical tools, and we just can’t go back to the way it was, now we’ve become sort of reliant on a lot of these different, additional value that these things provide. So I think what you’ll find is that it’ll necessarily accelerate that digital transformation or the remaining transformation, because we’ve come to rely on a lot of the value that the current digital transformation provides.
Jackie Whisman: Yeah, I think a lot of us were pleasantly surprised that our existing infrastructure was able to hold up with all of this new needs, but really, it does seem like we still have a long way to go.
Pat Combes: We do. Yeah, there’s a lot of transformation yet to happen, but I think the value that it’s provided, especially if you look in the current crisis with things like telemedicine, telehealth, remote visits, continuing care for people in clinical trials or things like that that can’t physically visit their provider, these are all things where we have the existing tools. And now that we’ve begun to move a lot of these processes into the digital space, so to speak, it turns out that they actually provide a lot better ways to connect and care for folks who are spread out over remote areas or are normally unable or for whom travel is very difficult or things like that. And as we move back, people aren’t necessarily going to want to just go back to in-person visits, now that they’ve counted on the value that remote visits can provide.
Jackie Whisman: Yeah. I never want to go see my colleagues in person again, that’s for sure.
Pat Combes: Well, we’ll get to there, but yeah.
Daniel Castro: COVID-19 has also, I think, revealed the importance of data, such as for modeling population health, sharing information across organizations, and measuring the effectiveness of treatments. What types of investments do you think we still need to make to strengthen this healthcare data infrastructure, not just for potential future pandemics, let’s hope there aren’t any more, but for other diseases and chronic medical conditions that are out there?
Pat Combes: I would characterize this a little more as that I think that the infrastructure as it stands right now or in the trajectory that it’s on is actually more than adequate for a lot of the things. And I think, like kind of what we just referred to in terms of its ability to absorb and provide for a lot of the digital needs of the current pandemic, really indicate that. I think instead, the way that healthcare data is collected needs to be a little bit reframed. Right now we really view things as a series of encounters. If you look at population health systems that are creating longitudinal records for people, it really is a sort of second order representation of that person. It’s reconstructed from encounter data that is taken over their lifetime, and it’s a lot like if you were to try and write a biography of somebody just based on their public appearances or public speeches or something. It’s going to be incomplete. It’s not a complete representation of that person.
And now we have a lot of the tools necessary. A lot of the tools, consumer wearables and things that exist, are capable of collecting detailed, accurate healthcare information about you, and yet there is not really a great way to represent that in a longitudinal view of that patient. And so I would say that what needs to be done is less in terms of altering the core data infrastructure so much as the data model layered on top of that infrastructure, where we can look toward building a better representation of a patient over their lifetime in order to provide them with better care, more timely care delivered, instead of just looking at things when patients are just encountering the healthcare system. And that’s usually when everybody’s at their worst, so it doesn’t give you a complete view of that person.
Jackie Whisman: Right. When they ask you to fax something, that’s when I lose it.
Pat Combes: Yeah. Yeah, and you’re already in a bad state, because you’re-
Jackie Whisman: Exactly.
Daniel Castro: I really like that concept we talk about, about getting a fuller representation of a person, and I think some of that will be coming, as you mentioned, from wearables, from these connected devices. Some of it also might be coming from other datasets that are key representative sources, whether it’s somebody’s educational or employment history, where they’ve lived, these types of important informational assets that can affect people’s health. How are we integrating more of those types of information into these records? Because we did a report at ITIF that tried to look at the type of advancements that are occurring in data-driven medicine, and what we saw is there are so many opportunities to improve this, whether you’re talking about using ML for looking at protein structures, doing clinical trials and you’re trying to identify the right patients, or in the delivery of healthcare. But all of it required access to much more information, much more data, lots of diverse sources. How do we start bringing all of that together?
Pat Combes: If you look now, I think the cloud is really powering a lot of the collection and the bringing all together of this data into a place where it’s all accessible. I think one of the big transformations that that needs to happen here is a better way to account for all the data, to measure the value of that data. And I don’t just mean it in the sort of master data management sense, but even beyond describing the value or potential of that data to be applied to different modeling or scenarios or patient needs or anything like that. Right now, we’re at a point where we can really pull everything together and pull a lot of different diverse data sources together into one place and power a lot of new analytics that way, but we need a better way to account for it or sort for it, through it.
I think of it a lot like a giant library. Right now, we can assemble a giant library of all the known, everything that has ever been published and everything that’s ever been written, and we can pull it all together in one place, but it might not do you a whole lot of good if you can’t find it, what’s inside. It just might be fun to hunt through for the rest of your life, but if you really want to make meaningful change or really address something, we’re going to need to find a better way to describe the significance of that information has gathered for different scenarios. That’s a work in progress, I think.
Jackie Whisman: What’s the technology you’re most excited about right now?
Pat Combes: Not to sound too cliche or anything, but I think it’s still AI, and really the ability for AI, not in the sense of making new advancements, though I think plenty of new insights are going to be generated in the future from AI, but really to make things more relatable and easier to use for everybody without having to rely on a small set of experts or anything like that, but being able to scale that expertise over the entire population.
Jackie Whisman: It’s almost like Daniel Castro wrote that answer, because he loves that stuff. He lives for accessibility.
Daniel Castro: Looking forward to the future.
Pat Combes: Yeah, I think it’s important. And I think about my grandparents or anything, where they’re bombarded with lots of information that may be hard to understand and really compile all at once. And what does it mean? What is the significance? And if you had something that was helping you sort through it, that’s the real power that AI can provide. And that’s an exciting thing in the future.
Daniel Castro: I think we’re getting closer and closer to that every day, in part thanks to the type of work you’re doing. So I really appreciate you sharing these thoughts with us.
Pat Combes: Sure thing.
Jackie Whisman: Yeah. Thank you, Pat. It was great to have you. And that is it for this week. If you liked our show, please be sure to rate us and feel free to email show ideas or questions to firstname.lastname@example.org. You can find the show notes and sign up for our weekly email newsletter on ITIF.org and follow us on Facebook, LinkedIn, and Twitter @ITIFdc.
Daniel Castro: That’s it for now, but we have more episodes coming.
Jackie Whisman: Come back.