Episode 56
HIMSS-22: Top Health IT Trends
with Kaveh Safaveh, M.D., J.D., John Glaser, Ph.D., and Suchi Saria, Ph.D.

March 29, 2022

Watch

Share on email
Share on twitter
Share on linkedin
Share on facebook

Listen

Kaveh Safavi, M.D., J.D.
Senior Managing Director, Accenture

Kaveh Safavi, M.D., J.D., is a senior managing director at Accenture where he is responsible for leading, developing and driving a growth strategy that differentiates Accenture’s offerings for providers, health insurers, and public and private health systems across the globe.

A seasoned executive, Dr. Safavi brings more than three decades of leadership experience to Accenture Health. Prior to joining Accenture in 2011, Dr. Safavi led Cisco’s global healthcare practice. Before that he was chief medical officer of Thomson Reuters’ health business, vice president of medical affairs at United Healthcare, and had leadership roles at HealthSpring and Humana. Among his many accomplishments was establishing one of the Midwest’s first electronic-health-record-enabled primary care practices.

Dr. Safavi has published numerous papers and is often quoted on healthcare issues in various media publications, including The Wall Street Journal, the BBC, The New York Times, Consumer Reports, US News and World Report, Harvard Business Review and The Economist. Recently, IT Services Report named him the #1 healthcare IT executive for 2020.

Dr. Safavi earned an M.D. from Loyola University School of Medicine and a J.D. from DePaul University College of Law. He is board-certified in internal medicine and pediatrics and completed his medical residency at the University of Michigan Medical Center. He serves on the Weinberg College of Arts and Sciences’ Board of Visitors–Northwestern University, is a frequent guest scholar at the Stanford University Clinical Excellence Research Center and serves on the advisory committee of the Buck Institute for Research on Aging.

John Glaser, Ph.D.
Executive-in-Residence, Harvard Medical School

John is an Executive-in-Residence at the Harvard Medical School Executive Education. He is a former senior vice president of population health, at Cerner Corporation. Previously he was chief executive officer of Siemens Heath Services. Prior to Siemens, John was chief information officer at Partners HealthCare.

John was the founding chair of the College of Healthcare Information Management Executives (CHIME) and the past-president of the Healthcare Information and Management Systems Society (HIMSS). He is the former Chair of the Global Agenda Council on Digital Health, World Economic Forum. He is former Senior Advisor to the US Office of the National Coordinator for Health Information Technology.

John serves on the boards of the Scottsdale Institute, National Committee for Quality Assurance, Wellsheet, and Relatient.

John has received numerous industry awards including the John P. Glaser Innovation Award (established by the School of Biomedical Informatics, University of Texas Health Sciences Center), William B. Stead Thought Leadership Award (American Medical Informatics Association) and the Lifetime Achievement Award (CHIME).

John received his Ph.D. from the University of Minnesota. He has written over 200 articles and three books on the strategic application of IT in health care. He is on the faculty of the Wharton School at the University of Pennsylvania, the UTHealth School of Biomedical Informatics and the Harvard School of Public Health.

Suchi Saria, Ph.D.
Founder and CEO, Bayesian Health

Suchi Saria, Ph.D., is the Founder and CEO of Bayesian Health, the John C. Malone Associate Professor of computer science, statistics, and health policy, and the Director of the Machine Learning and Healthcare Lab at Johns Hopkins University. She has published over 50 peer-reviewed articles with over 3000 citations and was recently described as “the future of 21st century medicine” by The Sloan Foundation. Her research has pioneered the development of next-generation diagnostic and treatment planning tools that use statistical machine learning methods to individualize care.

At Bayesian Health, Dr. Saria is leading the charge to unleash the full power of data to improve healthcare, unburdening caregivers and empowering them to save lives. Backed by 21 patents and peer-reviewed publications in leading technical and clinical journals, Bayesian leverages best-in-class machine learning and behavior change management expertise to help health organizations unlock improved patient care outcomes at scale by providing real-time precise, patient-specific, and actionable insights in the EMR.

Dr. Saria earned her M.Sc. and Ph.D. from Stanford University working with Professor Daphne Koller. She visited Harvard University as an NSF Computing Innovation Fellow and joined the Johns Hopkins faculty in 2012. Currently, Dr. Saria is serving as an advisor to the FDA on AI/machine learning.

We can't separate technology from business anymore. And because we can't separate it, everyone needs a level of technology literacy.

Transcript

[00:00:06] Gary Bisbee, Ph.D.: The Health Information Management and Systems Society melds healthcare and information technology in one very large setting.  We invited three healthcare and IT experts, Drs. Suchi Saria, Kaveh Safavi, and John Glaser to discuss healthcare and IT trends, highlights, and innovations coming out of HIMSS-22.   We explored health IT issues like interoperability, the cloud, remote monitoring, and what was new and not so new on the HIMSS floor.  One new topic was caregiver turnover and shortages which the panel thought were underrepresented. HIMSS showcased the cutting-edge applications of AI and automation. We discussed their use cases and the new challenges these technologies are facing, such as usability, privacy, security, regulations, and data bias. For all healthcare leaders the panel believes that it is fundamental to understand how technology is revolutionizing healthcare.

Well, good afternoon, John, and Suchi, and Kaveh. And welcome. We’re pleased to have you at this microphone and we’d love to get your views on the key takeaways and trends of HIMSS. We’ll talk a bit about a new meeting called ViVE later, but let’s focus on HIMSS now. And I asked earlier, but let’s go back. So when was the first HIMSS meeting that each of you went to? John, we’ll start with you.

[00:01:38] John Glaser, Ph.D.: Oh, it was 1989. And that was, at the time, the attendance at the conference was about 4,000 people. And we were thrilled that there was a 25% year over year increase.

[00:01:49] Gary Bisbee, Ph.D.: Yeah. And now it’s, I don’t know what it was this year. Suchi, they were saying, I guess, 26, 27,000. But it has been a lot more than that in the past. Suchi, when did you first start going?

[00:02:03] Suchi Saria, Ph.D.: So I went to my first HIMSS, I think, 2011.

[00:02:07] Gary Bisbee, Ph.D.: Okay. And Kaveh?

[00:02:09] Kaveh Safavi, M.D., J.D.: About 2002.

[00:02:11] Gary Bisbee, Ph.D.: Okay. So we’ve got a pretty good, I started right about the same time John did, by the way. Well, why don’t we start off with the value of HIMSS. It’s a large meeting focused on health IT. What do each of you find as the value of HIMSS when you attend? John let’s start off with you if we could.

[00:02:33] John Glaser, Ph.D.: Well, I think Gary, the value of HIMSS, much like the value of CHIME or the value of a number of associations is you bring the community together. And you cement relationships that are in place because, sometimes, this is the only time you see people is once a year. You form new relationships. I always think, well, if you pick up five or six pearls or insights along the way, you get feel for the floor and this, that, and the other. But it energizes the community and enables them to have a certain set of conversations. You know, 10 years ago, it was meaningful use. What’s the deal with that? This year, a lot of discussion on FHIR and AI, et cetera. But if nothing else, bringing the community together as a community to live and to learn and to reestablish relationships is always a very useful thing to do.

[00:03:14] Gary Bisbee, Ph.D.: Suchi, how do you think about

[00:03:15] John Glaser, Ph.D.: it?

[00:03:17] Suchi Saria, Ph.D.: With my background I used to go to, and I still do, go to a lot of academic meetings, academic meetings in MLAI, and academic meetings in medicine, moving to sort of more of an industry conference where I would mostly attend to give talks. I think this year was super interesting because I felt, as AIML is moving from research into the front line, I think five years ago, if I walked down one of these industry conferences, it felt like, I think, if I may say so, many experts would roll their eyes with hearing the things they were hearing in terms of marketing and I felt like it was really exciting to see more and more of ideas actually turning into practical things that people are starting to implement and adopt. So actually walking down the aisle, it was kind of a very beautiful, easy way to get a preview of what is actually happening in the practical realm. How are people adopting it? What are ways in which people are adopting it? How are they talking about it? Where are the gaps? So in two days, you could kind of get a preview extremely fast.

[00:04:19] Gary Bisbee, Ph.D.: Kaveh, we’ll let you bat clean up here. How do you think about the value of a HIMSS meeting?

[00:04:24] Kaveh Safavi, M.D., J.D.: For the most part, I’ve gone really through the view of a sponsor and a company for the trade show. So for me, it’s predominantly a trade show. I’ve been going for 20 years. By the time I started going, I already had some knowledge of the sector, but I did take advantage, personal advantage, of the learnings. But I’d say, for the last, at least, decade, my view has largely been about how you would approach an event like that to meet your trading partners. And as a trade show, you spend more of your time focusing on selling partners rather than buying partners. At least that’s the experience that I have, and that continues to be the case.

[00:05:02] Gary Bisbee, Ph.D.: Let’s look at this a bit differently. We had the HITECH Act pass, roughly, what, 13 years ago, 2009. John mentioned meaningful use. But what contribution has HIMSS made to the field of health IT, do you think, Kaveh?

[00:05:19] Kaveh Safavi, M.D., J.D.: Well, again, from the perspective that I take, it is without any question the single largest healthcare IT trade show. And to the extent that, in any kind of an industry where you have that kind of adoption, every one of those industries needs a place for convening, not because you can’t go out of your way to make individual meetings, but because the idea of having everyone together at one place creates an opportunity, not only for planned, but unplanned interactions, which become the basis for partner relationships. I think it’s critical. And because we’re moving into an environment where no single entity has the answer, the technology solutions are increasingly ecosystem and partner driven. So the value of places to find partners just goes up.

[00:06:06] Gary Bisbee, Ph.D.: Suchi, how about you? You’re on a cutting edge in a lot of the work you do, so you might look at HIMSS a bit differently. But, what has HIMSS contributed to the field?

[00:06:19] Suchi Saria, Ph.D.: Yeah, I think, like Kaveh said, the fact that so many different people are coming together and learning each other’s language is so key. And we need more of that. A little bit of a flip point of view. I felt like the size of the conference did feel quite overwhelming. It was smaller this year than past years, but what that meant is it felt a little bit like a zoo. And you know, you really sort of had to come in with an agenda of what you’re hoping to accomplish because there’s so much going on. For example, I spent a lot of time at the AIML forum. Maybe it was about athousand people roughly. I had a chance to give a talk at the AIML forum, which was super fun and really got very engaging questions. And one thing that I really loved about it was, compared to other meetings, this particular one felt like people were just much further. They’d worked on ideas, they’d implemented ideas, they understood where things break, and they were very prepared to hear about how to go about tackling these issues. And they just felt like they were very grounded and not just theoretical. So it was really, really fun to be able to engage with that audience. So I think that breakout group was great. And so the extent to which there can be more breakout groups always helps, but it did feel quite, I mean, there were people I wanted to meet, we met, but it just felt, as a meeting, extremely large. I’m curious to hear what John thinks because, you know, he was on the HIMSS board. So this is in a way your baby, to some extent, growing to the size it has. What do you think?

[00:07:51] John Glaser, Ph.D.: Well, I mean, at its peak was in the low forties, 40,000, and it was overwhelming. It was a zoo, you know, trying to get a cab to get to the restaurant. Good luck, in a variety of ways after certain meetings, et cetera. So, and I think, you know, clearly it’s added to the, industry. It’s kind of hard to know how much. Different meetings have different purposes. You go to CHIME, it’s much more intimate. It’s with people who are your colleagues or CIOs or CFOs, or CMOs, or whatever. You go to AMIA for the informatics, you know, the more academic sort of… Suchi, you have certain things you go to in the ML community, which are targeted, et cetera. This is the big tent. And I think you can always advance the cause. You have to be focused, to Suchi’s point, about who you’re going to meet with, and Kaveh’s point, and why, et cetera. But nonetheless, people are learning from each other. People are forming partnerships and relationships, the kind of glue in the engine that makes things happen and across the board. So, it’s kind of hard to put a sort of definitive figure other than, you know, in the course of major undertakings of moving an industry, it takes a lot of people working together and going in the same direction and teaching each other. And this is a sort of agar, for lack of a better word, that enables all this stuff. So I always thought there’s enormous power bringing people together, a little bit of structure, and let them mingle. And they’ll do amazing things.

[00:09:02] Gary Bisbee, Ph.D.: Well, the COVID question is, how did it feel to be with 20,000 of your closest friends?

[00:09:08] Suchi Saria, Ph.D.: I’ll actually take that one because it’s super funny. I was in Florida for a couple of meetings right before that, including ViVE. I don’t know if it had something to do with the pollen or something else, but I definitely triggered my allergy and I was coughing a little bit. And I felt terrible being in the room coughing, where I’m pretty sure people, you know, I had taken the negative COVID test and everything, but that doesn’t preclude, in that meeting, when you’re sitting at the table and you have to cough, even with your mask on. I was definitely feeling bad for the people around me.

[00:09:39] John Glaser, Ph.D.: You’re probably less likely to get a screwy, eyeball than if you smoked a cigarette in the middle of a meeting, you know. Both of those, you’re a leper.

[00:09:49] Gary Bisbee, Ph.D.: Kaveh, is that the largest meaning you’ve been to since COVID?

[00:09:52] Kaveh Safavi, M.D., J.D.: I was at Mobile World Congress in Barcelona two weeks ago. It’s normally 120,000. It was a 50,000 person meeting. But we had masks at Mobile World and now we’re mask optional. So I would say it was about as normal sort of a pre-2020 event as I’ve been to it since March of 2020. And I would say that there was a very clear sense from most people that they wanted to be in a congregate setting. For my own practice, which is highly distributed, I have thousands of people that are never in the same place at the same time, these meetings have always been an essential part of our familial relationship. And for us, it was like homecoming. I mean, we went out of our ways to spend time with each other and go out to dinner together, just us, because it was a family reunion. We’d only seen each other on camera for two years.

[00:10:45] Gary Bisbee, Ph.D.: Very cool. Well, what about trends? So what are the trends that you all picked up? Were there any new trends or was it just existing trends that we’ve been familiar with over the last several years?

[00:10:58] Kaveh Safavi, M.D., J.D.: I’m happy to just sort of give a couple of thoughts on this. The first one is that, apparently everybody uses the same people for their marketing because, if you just walk down the middle of the aisle and you look, the same four words show up on everybody’s list. What’s quite funny actually is how the same words can appear in branding statements all over the place. So. I understand in a way why that is, but when you walk down and you see them all in front of you at once, you realize that the idea of being distinctive is actually something that probably deserves a little attention rather than using words that you’re used to hearing other people use. One of the things I think that’s interesting about HIMSS is that, by its very nature, it tends to focus on what we’ve typically described as mature technologies. We think about technologies via an S-curve and most of the real money is spent on mature technologies, but a lot of the interesting conversations occur on a sending limit, the S-curve. And then the really interesting ones at the bottom of the S-curve. At HIMSS, it’s always been about sponsors and companies. So it tends to live at the top end of the S-curve. The startups are sort of all the way over in one corner and not really central. And so people either seek that or don’t see it, which is interesting. But I think something really interesting about this HIMSS, and it has nothing to do with HIMSS, it has to do with timing, which is that HIMSS is planned out a year, whatever, in advance, But for healthcare, and particularly healthcare delivery, the rapid ascendancy of staffing as a problem, and the staffing shortage is a problem, and healthcare costs related to payroll cost, has moved up the list of priorities ahead of marketing and ahead of assets. So while you see a lot of technology that talks about how to make healthcare better, when you have conversations with people about the problems they want to solve, it’s actually, how do we sell for the fact that we don’t have enough people to do the work? That’s not what the products and the booths are organized around right now. And I think that to me was most interesting and something I share with my colleagues, which is that the topic de jour is moving pretty fast right now because of context.

[00:13:05] Suchi Saria, Ph.D.: I’ll add to what Kaveh said. Actually, I totally agree with him around, it was, again, going down the exhibit hall, it felt like it was nuts. How much money had gone into those booths? Those, you know, like flying saucers everywhere and all sorts of things to be able to get attention and to stand out. One thing I empathized with deeply was I was trying to think about, you know, again, from having worked in the field for a long time and walking down the aisle, I was like, oh my gosh. I work a lot with Chief Clinical Officers, Chief Medical Officers, Chief Quality Officers, and CIOs, and I was walking down the aisle and I was like, if I was in their shoes walking down, holey majoley, I would be overwhelmed because it’s so unclear what is real, what is not. And then the other piece I was very pleasantly, you know, this has been sort of a little bit in the last couple of years already starting to bubble up and because staffing being such an issue, the importance of augmentation, the importance of automation. So I felt like I saw much more real emphasis, and maybe my eyes go there more quickly, but much more real emphasis on AI, not in a hypey sense alone. I mean, of course, when you’re in a vendor meeting, there’s going to be hype everywhere. But going beyond the hype, there were also very real ideas and real results and real, like there were good groups doing high-quality things, which was very exciting to see.

[00:14:26] John Glaser, Ph.D.: I think it’s hard. I mean, one of the things that HIMSS has done for years is they poll attendees about top 10 issues. What’s interesting is, year over year, those issues move all over the place. So usability is high one year, drops down here and you think, golly, how could this be? Like we solved it last year and so we no longer need to talk about it? And so part of it’s a little fickle in some ways. But nonetheless, I think you can see a mixture of things. So you see one year you’ll see issues that just really weren’t there the year before. I mean, so I go back to meaningful use in 2008. It wasn’t there, but 2009, it sure was. This year, health equity and digital equity is clearly on the map. It was not there, you know, last year or the year before. So also things are merged for whatever set of reasons. Similarly, you get these kinds of bubbling issues that all of a sudden take off like tele-fill in the blank. Telework, telehealth has been around, but all of a sudden it really took off, et cetera. And then you see, to Suchi’s point, the steady maturation of AI and ML. It just keeps maturing along the way. So you get this kind of interesting mixture of things where you said, I could see it five years ago, it’s just further along, and other stuff that says, I could kind of a glimmer five years ago, but man, we’re on a kind of an S-curve. And the other kind of stuff is, where in the world this come from? Now, you know where it came from, but all of a sudden it’s there. You know, one of the things that was kind of a, really, a lot of attention this year was interoperability, FHIR, because there’s the FHIR mandate by the end of the year. And all of a sudden the information blocking rule is kicking in here and all of a sudden, TEFCA, you know, that sort of way we’re going to get the network of networks. I think sometimes in healthcare, we wait until the last minute to do something. And all of a sudden, it’s the last minute. And so there’s a lot of anxiety about this. So anyway you get this kind of mixture of issues popping up and going away and setting a picture, et cetera.

[00:16:08] Suchi Saria, Ph.D.: Yeah. Like John said, there was a lot of word toolkit discussion, which was interesting. And it was kind of fun because there’s a little bit of, like, in the moment when you’re responding, it’s much easier to do that with marketing. It’s harder to do that with product. It takes a long time to build the product to work correctly. It’s much easier to put in a nice glossy white paper to talk about it. So there was definitely a lot of glossy on, like, virtual care and how virtual care could look. I don’t know if you’re seeing the right virtual care of the future today, but we’re definitely seeing a lot of glossy on it.

[00:16:36] Gary Bisbee, Ph.D.: If AI was just a hundred percent diffuse throughout, that would be some point in the future. What percentage would we be at now, 10%, 20%, 30%?

[00:16:48] Suchi Saria, Ph.D.: Yeah, I think what I’m starting to see at least is maybe like pretty much across the continuum, like examples all the way from like pathways, clinical trial recruitment, falls detection, clinical augmentation, earlier stages like falls, sepsis, areas where I work a lot in, administrative cases, billing, virtual care, at home care. So I feel like we’ve gone from sort of hypothetically talking about it to more results. And I’m seeing early results in a number of these areas. In some cases we’re starting to now see these early results meet across systems. And I generally say, because of the staffing shortage issue over the last year, my sense is, like John said, some things rise very quickly through the S-curve. If we can succeed in demonstrating that we actually can tangibly solve for augmentation effectively with AIML deployed correctly, I think that’s going to be an example of a place where you’ll see it flying through the S-curve very, very fast. But yeah, it did feel like 2030 is maybe what I’d say where we are. We’re not at zero. We’re not at five. We’re not at 10. I think there’s great desire to help understand how to implement it. You know, many groups are attempting to work in it, but not as many groups with deep expertise.

[00:18:13] John Glaser, Ph.D.: I’ll just defer to Suchi and Kaveh on this. I mean, I think it’s a hard question to answer because there are parts where it’s really far along. I mean, golly, the voice recognition, and checking emotion, and also, you know, facial recognition. Yeah, I get the bias thing, but still, you know, it was pretty far along or, you know, my old Siemens days, you have the AI and the equipment that says, by the way, this thing’s about ready to go south unless you do something. Or at a lot of administrations, staffing, prior authorization, et cetera said, wow, we’re making some pretty good progress here. On the other hand, you say to what degree are docs using it daily, and of course the clinical practice. Well, maybe not. So it’s a little uneven across the board. And I also think, you know, at times, we have to remember you know, an algorithm is not a solution, which is not a company. And so you get a lot of algorithms out there. Pretty cool. Now whether they’ve been tested well, separate question, et cetera, but you can get a lot of energy on, I did this algorithm and it looks pretty damn slick, but that doesn’t mean you got a solution here. And that doesn’t mean you got a company. So you can say how far along in the algorithm? Well, do a lot of cool things. How far along are the companies? Well, there’s a funnel going on here. And then, th last point, and then turn over to my colleagues who know more than I do. I always sometimes think that nobody buys AI. They buy something else that performs better because of AI. And so they buy a revenue cycle system that is smarter because of AI or they buy a, you know, imaging system. So the real question is, well, I’m buying, you know, imaging capabilities or revenue cycle, and you’ve got AI. Why is your product better because of it? So, anyway, that’s what you buy. You buy a car that’s safer because of AI.

[00:19:41] Gary Bisbee, Ph.D.: Kaveh, what do you think?

[00:19:42] Kaveh Safavi, M.D., J.D.: I think we’re in the same zone here, Suchi. I was saying I was going to say, second or third inning in a nine inning game, so on the same order of magnitude. I think about it largely as particularly machine learning. You know, it’s pretty pervasive as an analytic capability now. That’s kind of the no part of the world, but there’s a real opportunity is really in the act and the do part of the world. It hasn’t really made its way much into workflow and workflow substitution. I think there’s huge amount of opportunity there. And it’s going to first show up on the cognitive side. And I think we’ve wildly underestimated its value around the physical side of work, because there’s a lot of task substitution that needs to occur on the physical side. So people are aware of it, but I think we are still, there’s still a way, a lot of distance between the full benefit we’re going to get. So, opportunity.

[00:20:31] Gary Bisbee, Ph.D.: We’ve got administrative uses use for, you know, finance and so on. But Suchi, could you share with us your thoughts about the clinical uses of AI, which may be in a different place than some of the administrative uses?

[00:20:47] Suchi Saria, Ph.D.: Yeah, absolutely. I think, like John said, right, people are buying something because of a problem you’re solving rather than the benefit. What are the problems we’re seeing now, right? So, a huge aspect of it is, like last year, we saw, you know, staffing shortages coupled with significant drops in quality, safety. You know, outcomes plummeted last year. What we’re also going to see as a result is risk. We’re also seeing a lot of emphasis continuing on that side, continuing, not a sudden surprise, but like a shift towards pop health. So what that means is my sense is, you know, things like patients at risk for adverse events that we can monitor and model and make it easy for a small set of teams to be augmented to quickly respond. You know, managing at risk contracts more confidently because you can identify patients who are at risk for getting admitted or readmitted and doing something about it. More patients are at home, or, you know, there’s a little bit more patients who were in the hospital and now heading home and maybe previously they would have gotten the care inside the hospital. They’re sort of starting to get more of that care at home. Naturally, the need is to stitch together that siloed, you know, these two separate experiences into a single episode to start thinking about how to do that transition more effectively, but also to more seamlessly monitor because you can’t have a full team at home in the same way you have them while you’re in the hospital. So, I see a number of use cases, which, pre-COVID were very much nice to haves and not something people were thinking about as hard, but now feels has become a very real thing. The extent to which a health system’s already engaging and doing something about it, I think is a secondary story, right? So I think this is where John and Kaveh were saying. My sense is, over the next couple of years, we will see how much this becomes a priority. How much are they really deeply engaging to actually do things differently as opposed to business as usual?

[00:22:46] Gary Bisbee, Ph.D.: What about data? John mentioned we’ve got interoperability, some new standards at the end of the year. We’ve got data blocking as an issue it’s being dealt with to some degree, at least. And obviously, AI, machine learning, required data. So will these changes coming down the pike over the next 12 months or so, is that going to free up data and make more data available for the use of AI?

[00:23:15] Kaveh Safavi, M.D., J.D.: Well, so lots of, certainly, I mean, there’s a lot of raw materials. For it to be useful, there are a few things that have to happen. Like right now, one of the challenges we have, which we have to get our arms around it, is, just because the data exists doesn’t mean you’re permitted to use it. And so dealing with privacy and consent is a non-trivial issue. That’s a governance and a policy issue, not a technology issue. First, becomes a technology issue. But it has an element there that’s important. We understand that, for AI to be trusted, one of the predicates is that the data cannot be biased. Well, the data that we have available is largely either a reflection of the way we do things or a sample that’s available, both of which have bias associated with them. So we want a machine to make decisions based on what we put in and, by its very nature, it’s biased. We build in a lack of reliability in our system. So we’re going to need to solve for those problems. Increasingly it’s less about the quantity and it’s more about the heterogeneity, as well as chronology. And so, for really interesting data on healthcare around a person, we have to get a complete picture of them, which won’t come just from their clinical record.

[00:24:26] Suchi Saria, Ph.D.: I want to add to what Kaveh said. Actually, this is one of my favorite topics. I’ve spent a ton of time doing research in this and publishing in it. I think there’s this misconception. If you think about it as innings, if you will, initially I think we felt, oh my gosh, can we get the data? I think then the data became available. Then people started to think, oh my God, we need a lot of data to be able to do it. Then the next state was realizing, oh my gosh, when we get a ton of data and we blindly just started putting black box methods, we end up getting methods that are not all that reliable. I’d say the next inning, and again, in research, we’re further along than industry practices. The next inning in research, which is almost four or five years ago, was sort of recognizing that we need techniques that can really correct for these biases. So I’d say, on the research side and state of the art side, we actually have a fair number of techniques that know how to think about how to learn models that are reliable, how to measure for bias, how to monitor for bias, how to avoid solutions that create bias. But I wouldn’t say all of these ideas have migrated widely into industry where, you know, we’re seeing a lot more, there’s a very variable implementation of tools. And I think around this, we just sort of, yesterday in STAT, wrote an article, this is a new coalition we formed with a group of colleagues around trying to make some of these best practices clear. In my work with the FDA, we’ve grabbed some of these tools from the field and something they can use for starting to make it easier to monitor and evaluate efficacy, right? And we use a lot of these ideas within the company, Bayesian Health that I founded, but I definitely see that knowledge of these best practices are still very sparse and something we’re hoping as more, you know, the healthcare coalition we’re part of, but like as other groups come along, I think these best practices will start to become more visible and hopefully more uniformly implemented.

[00:26:21] John Glaser, Ph.D.: Gary, just to add to this, I think, one of the things we get into this conversation every now and then is sort of the truth that data is data, sort of this homogenous, all the same., It ain’t all the same. And in fact, particularly, depending on your use, you’ve got the data you need now. You don’t need any further interoperability if you’re looking at predicting someone in the ICU, whether they’re going to head south or not, et cetera. So I think you’ve gotta be, it depends on your purpose, you know, whether or not there’s a lot of data you’re going to have, or whether it’s good or not good, et cetera. I do think, on the topic of interoperability, and a number of things are happening this year, which, from the federal level, will position us to make products, whether it’s the adoption of FHIR. We’ll beat you up if you don’t. Or TEFCA or the blocking rule, et cetera. And we’ll see in the years to come how much the industry actually leverages that because, actually, interoperability is a thing you decide to do. You know, you decide to share data. It’s not something that happens automatically because you got APIs, etc. I remember a couple, two years ago, actually, looking at, well, what can we learn from banking or travel about interoperability. They’re further along. And one of things you learn is no industry is fully interoperable, not at all, et cetera. But where are they are, they do three things. They focus on a narrow set of transactions, which have value to all. So the SWIFT, which has come up in this sort of Russian thing now, is sort of the exchange between banks. There’s real value there. The second is where there’s a real business case. Everybody’s sitting around the table saying, I know what’s in it for me and I’m willing to pony up money and, you know, time and effort to make it happen. And the third is where there’s a buyer that convenes everybody to hammer through the issues. So we’ll see to what degree we, as an industry, we’re not going to share everything. It’s never going to happen. But where will we focus and how valuable will that be? And can we, as an industry, agree to go after prior authorizations or other types of things that, we go after. So we’ll see. We’ll see whether we mature along the lines that other industries have done. And we’ll be further along. Whether it’ll be further along to everybody’s purpose because of what they’re trying to do for their algorithm or this, that, and the other, we’ll see. We’ll see.

[00:28:15] Gary Bisbee, Ph.D.: You know the question, John, just to follow up on your point is, the government now, federal government, is not only the largest payer, but the largest regulator. And so that progress you’re talking about seems to be heavily related to how capable the government is to move quickly or expeditiously. Is that right? John? Am I thinking about that the right way?

[00:28:41] John Glaser, Ph.D.: Well… well, I think, clearly, you pick any country you want. The government is the most potent actor in healthcare because they’re the largest employer, and so they’re taking care of their folks. They’re the largest provider of care. They’re the largest purchaser of care. And then they have this role called, you know, protector of the common good when the free market fails. You know, so they’ll step in and do certain things like that. Spending a year at ONC to help put together the meaningful use regulations and what was interesting to me is it has such power, you know, because of what the rules said. I mean, it’s unbelievable. And so one of the things that you have to do when you write the regulations, and the regulations take the law and translate it into, well, here’s what you got to do. Let’s get real. What does meaningful use really mean, et cetera. They have to send these things out for comment. You know, they have to listen to the comments. They have lots of groups who come in and say, by the way, it should say this. It shouldn’t say that if this, that and the other. And so they have to because it’s government and you have to listen to your people, et cetera. And you have to do it because it’s called risk mitigation, because it’s so easy to get it wrong. And man, it is hard to change regulation. You can really screw things up if you get it wrong. So you better subject it to lots of other pairs of eyes. But golly, Gary, that takes time. You know, to write the rule, to put it out for comment, to entertain everybody, this, that, and the other. And then, it takes time to figure out, well, we messed it up. We should have had a comma here, but you didn’t. And then it takes time to figure that out. And then it takes time to revamp the whole thing. So because of its power, it’s got this very deliberate risk management approach and it’s going to be slow. And it will never have the sort of, you know, like Suchi and, or, you know,Kaveh and their companies can move pretty darn fast because they don’t have to deal with all this stuff. So, a long winded way of saying, you know, it’s got power. It’s got to be very judicious and thoughtful about the use of that power, which means it’s going to be deliberate.

[00:30:25] Gary Bisbee, Ph.D.: That’s well said. Let’s turn to the EMR companies. Any advancements or new news out of the Epics and Cerners and Meditechs of the world?

[00:30:36] Kaveh Safavi, M.D., J.D.: Well, I think that the topic du jour right now is the idea that people are going to actually host their EMRs on public clouds. And what’s really driving it as they’re trying to move their overall technology estate to a public cloud for primarily around resilience and cybersecurity and pace and their ability to innovate and connect to other partners. So, not EMR per se, but because they need to do it. And if you’re a healthcare provider and your EMR is not on a public cloud, you haven’t moved to the public cloud enough. And so you see a lot of that dialogue and how that’s going to play itself out at. And it’s confounded by concepts of sovereignity globally. The lack of willingness of non-US countries to do business with US cloud providers is a complexity that is emerging and it’s only heightened by geopolitical things. So, you know, just because it’s possible doesn’t mean it’s permissible or doesn’t mean it’s politically viable. So we’ve got a lot of moving parts with respect to that, but I think that’s kind of the topic du jour right now, or at least one of them.

[00:31:38] Gary Bisbee, Ph.D.: John and I are close to Cerner in different ways, each of us, but what’s the scuttlebutt on the Oracle acquisition of Cerner? Any talking about that on the floor?

[00:31:50] John Glaser, Ph.D.: Oh, yeah, sure. I mean, because it’s big news and there’s, you know, sort of everybody’s speculating about this, that or the other. I don’t know how well informed it is, Gary, but shoot, that shouldn’t stop It’s like speculating on The Bachelor, you know, whether he’s going to make a mutton head choice or not making a mutton head choice. And it turns out he made a mutton head choice. But anyway, there was a lot of speculation about all that. So I think my sense is, you know, that’s obviously a topic. What does it mean? Where will it go? There’s also the topic of Meditech and Google doing a partnership of some of the Google technology within Meditech, et cetera. I think, frankly, the more interesting conversation, at least for me, and I’m not sure how much of it is really a floor conversation, is I think progressively what we’ll see Gary is a series, a sort of ecosystem of apps that surround the EHR, leveraging FHIR, and this, that, and the other. It’s like what you see in banking. The core transaction system remains. And it doesn’t mean that Epic and Cerner don’t innovate. They continue to do that. But you’ll see companies wrap around this kind of stuff. And that’ll be an interesting phenomenon to see how well that really works and how vibrant is that ecosystem really in making this thing happen because you know Cerner as well as I do, Gary, probably better. There’s only so much innovation it can do at a particular level and it’s the same with Epic. So I think the sort of broad trend is, if you take the exhibit hall, is how many people are actually able to make a real goal of it, you know, surrounding and sitting on top of Epic or Cerner or Meditech or eClinicalWorks, et cetera. And to what degree do the doctors and the nurses say, man, this sums up, this works pretty well.

[00:33:16] Suchi Saria, Ph.D.: I think for that to happen, John, one of the things that’d be really, really crucial is for EMR companies to evolve their identity from being like the solution that they’re competing for headspace to becoming enablers, right, whether they’re seeing themselves as the information superhighway on which they can like enable really, really fast high-quality traffic. I think that is an identity. You know, we’ll see if we actually, you know, get there in healthcare. I think it’s very important. It’ll move. It’ll be exciting. And I think I’d even say on the health systems, like, you know, health systems, payers, I think they’re at different ends of this journey of like how they view it, the extent to which EMRs as like the solution to all problems versus sort of their base layer infrastructure on which they will now start building an ecosystem of solutions to solve their problems.

[00:34:08] John Glaser, Ph.D.: I I’m sure that Kaveh sees this, but I think two things will sort of force the issue. You know, one of which is the fed saying, you got to make it easier, man. And if you start throwing, you know, things in the way here, we’ll come beat you up, we’ll penalize you. So the information blocking, FHIR, that will just make it easier. The second is one of the things that I find, and, I think, Gary, Kaveh, you guys probably see this, is the health system is getting bigger. All the big players are getting bigger. And so what happens when you get bigger is you have the resources, the ability, and the belief, I can do my own thing. I’m not wholly relying on Cerner and I’m not wholly relying on Epic. It’s different if I’m a 200 bed hospital. You got me. Okay. So I’m quite willing to say, here’s the deal. You got to open it up because I want to do A, B and C and Bisbee will because, golly, you’re a big customer, et cetera. So we’ll see very sophisticated digital health IT, people demand it, and we’ll see the companies respond because that’s what they do in these things. And I don’t know, Kaveh may see this or not see this, but I suspect we’ll get a more demanding customer.

[00:35:04] Gary Bisbee, Ph.D.: Let’s move to ViVE just briefly. Suchi, you were there. It’s a new organization focused on digital health, it seems to me mostly, I think. But it has vendors there and it has users of one kind or another. Suchi, how did you find ViVE? Give us a quick summary, if you could.

[00:35:24] Suchi Saria, Ph.D.: Yeah. So I think this was my first time attending ViVE. I don’t know if it’s the first ViVE meeting. I think it may be. It was way smaller than HIMSS. So I don’t know what the numbers were, but it seemed like maybe closer to 4,000, 5,000. In-person, it hardly matters, though, because you’re seeing people around you at 5,000. You’re seeing people around you at 26,000. It feels the same. There were some very marked differences in style. So, you know, ViVE was very much a, you know, let’s start with the look and feel, like beautiful illustrations everywhere, a great DJ that set the mood the entire time of the meeting. And so that sort of made it very like fun, easy to be here, easy to mingle. In terms of content, I felt like ViVE was much, much more positioning itself as sort of here’s where the field is headed. So a little bit more, like so many more startups were there. Many more talking about kind of the, little bit less like vendor booth heavy and more conversation heavy and more sort of let’s interact to like try to solve problems together, I felt.

[00:36:34] Gary Bisbee, Ph.D.: I think that’s a good summary. This has been an excellent interview. We really appreciate it. I’ve got one last question I’d like to ask each of you. And that is, for the up and coming leaders that are watching the show, what’s your advice for how familiar they should become with IT and digital health?

[00:36:55] Suchi Saria, Ph.D.: To me, it feels like the next 10 years is such an exciting time. The IT has gone from just being plumbing, like basic, basic plumbing in healthcare for it to become the catalyst for really improving healthcare delivery and the business of healthcare delivery. So I’m hopeful, you know, this isn’t, IT is no longer just a CIO issue. It really is an across the board care delivery, you know, enabler and people are educating themselves across the board and we’ll start to see companies at all sizes and scopes partner to be able to improve health outcomes, make healthcare more effective, more efficient, more competent, you know,

[00:37:34] Kaveh Safavi, M.D., J.D.: I think that we can’t separate technology from business anymore. And because we can’t separate it, everyone needs a level of technology literacy. We talk about this professionally now, but more importantly you’re talking about young people, which talk about this as a group and what we want our kids to learn in college regardless of their field. And I think we’re getting to a place now where everybody has to have an understanding of how technology affects a business that is more than just simply the label, but actually have some understanding. They don’t have to be coders. But I do believe that we’re at a place now where you can’t either be a leader or a participant if you don’t have, in our company, we call it a technology quotient, or some version of technology literacy.

[00:38:22] John Glaser, Ph.D.: You know, Gary, there’s a study done, actually it came out last year, by MIT looking at the impact of having digitally savvy leadership on transformation efforts. So they looked across a range of companies. And, you know, some of it was, well, what do you mean by digitally savvy? So, well, three things, one of which is they had a reasonable working level of knowledge about the AI. They could give you a credible description of AI. They could give you a credible description of blocking. They had enough, okay, they’re not going to go toe to toe. Second is, they had had scar tissue from a major implementation. They knew what it meant to roll out an EHR revenue cycle and, you know, how hard that is, this, that, and the other. Third is they had some experience with sort of piloting digital technology. Let’s go try it. And so, you know, good discipline and innovation matters. So if you had those three, what you find is that they actually had a remarkable effect, surprise, surprise, on how well the digital transformation of the organization was going. Now that assumes digital transformation is relevant. And I think in healthcare, it is. In organized religion, not so much. But you know, nonetheless, it assumes relevant, and it was not only the savviness of the leadership, it was certain practices they brought with them because of that. So they had an active, ongoing dialogue with the IT people where they were treated as peers, it was iterative, just nonstop, et cetera. They had a great way of sort of being learning. So they were open to trial. They were opening to minimal viable products. And they made sure that people were out in the industry, like these conferences, and really learning and sharing and being disciplined about learning. So actually, where’s the evidence? Not just, golly, I talked to this person and Chuck believes this is going to happen. So Chuck knows everything. So they wouldn’t go off to do that. And they were pretty modular. They could actually move people around. Golly, the winds are going this way versus that we’re going to redirect the budget. We’re going to redirect the team in the way we go here. And the other thing, Gary, is they sort of push the responsibility down to the trenches. You all in the trenches, you’re leading this thing. You know what the care is like. You know how to make it happen. I’m going to resource you. I’m going to coach you. You know, if you step into crap, I’m going to get you out of there, et cetera. But I’m also holding you accountable for getting the results here. So what you find is, to your point, yes, it matters. And there are certain specifics to it, but, more than you being the chief smart person who’s now a digitally savvy, you bring certain practices with you and you do the best that you can to get those across the organization.

[00:40:37] Kaveh Safavi, M.D., J.D.: Can I add Gary that if you believe in the metaverse, it’s going to matter for organized religion as well at some point.

[00:40:44] John Glaser, Ph.D.: I know, I ain’t even thought about that. Son of a gun. Like the psychedelic medication, I’m all for that, you know, I’m all for it. Timothy Leary was right. I knew it. I knew it.

[00:40:54] Gary Bisbee, Ph.D.: Great way to end. This has been a terrific panel. Thank you again. And by the way, why don’t we do it again next year after HIMSS and see what’s changed? Thank you all.

[00:41:04] John Glaser, Ph.D.: Alright, thanks Gary. Thanks. Thanks. Got it.

[00:41:06] Suchi Saria, Ph.D.: Thank you so much.

Subscribe for Updates​

For exclusive access to Think Medium content and program updates, subscribe here.