This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.

 All right. Welcome to this week in Health It where we discuss the news, information and emerging thought with leaders from across the healthcare industry. My name is Bill Russell, recovering healthcare, c I o, and creator of this week in a podcast. Dedicated to training the next generation of health IT leaders.

This podcast is brought to you by health lyrics, helping you build agile, efficient, and effective health. It Let's talk visit health lyrics.com to schedule your free consultation. If you missed it, we did 13, 10 minute interviews from the show. Uh, from the HIMSS showroom floor, uh, women in Health it with S Shade, Ann Weiler shared, uh, results from a Boston University study that showed better outcomes through digital engagement.

Uh, Joe Petro c t o of nuance showed us a clinic visit with no keyboard in the room, and that's welcomed. Uh, we talked. Policy with Anise Chopra. Consumer experience with Andy Crowder from Scripps Identity and Perimeter with West Wright from Impravada, whole Person Care with John Glasser and several others.

I hope you enjoy these. Check those out on the website this week in health it.com or the YouTube channel as well. Our guest today is John Halamka, the The Geek Doctor, as your book calls you, I'm not calling names here.

I'm happy to be here. And remember, geek is a term of affection Yes. To to, yes. To the technology, uh, class. It absolutely is. So you were really busy at himss. In fact, uh, we were both there for several days. I didn't see you once you, I mean, it's 45,000 people. It's, it's really spread out. Um, what, what were some of the things that you were, uh, you were doing while you were down there?

Sure. Well, I had 35 meetings and 12 speaking engagements in two and a half days, and I covered 75,000 steps. Wow. And so, you know, ch people actually said, I saw you everywhere. And I said, like, how is that? Well, because I was also in several videos that were playing in several places on the floor, and then I was, you know, screens and all the rest.

So it was a wonderful experience to actually be in three or four places at once, which is what I've always wanted to do. Uh, so, so what did I see at hims? Uh, so at HIMSS this year, the theme of course was patient-directed data exchange, right? Every HIMSS has a theme, you know, in the past it's been big data or cloud, and even last year it was a little blockchain, not so much this year.

This year it's with Sema Verma and Anish and Mike Levitt talk and Karen Salvo talking about some of the ways that the c m s and O C rules are going to empower patients with, uh, claims data and clinical data. You know, really kind of pushing us to, um, a future where instead of provider, provider data exchange, it's provider, patient, provider data exchange, solving a variety of privacy dilemmas.

Of course, we also saw quite a lot of APIs and fire and then every other booth. Machine learning AI was big. So those are sort of your big three themes this year. Big three themes. All right, so let's go back and forth on some of these. So, um, let's start with, so the, the federal government really has taken the center stage, literally.

I mean, they, they took up, uh, a lot of space with their announcement, dropping it right before the show. Um, Uh, happened. There was a lot of talk on the floor about, uh, transparency, data blocking, information sharing, uh, a patient centered, patient access, as you say. And in that keynote, what we heard is they're really trying to restore market forces to an industry.

Um, You know, that's almost consolidating to the point of opoly status. You know, you have, you definitely have it on the payer side. There's, there's, the payers have really consolidated. Now you're starting to see it on the provider side of which I've lived through it, and you're living through it, where you're at.

And um, and, and some would argue on the E h R side as well. So, um, You know, do these, do these, uh, rules that have just dropped, do they help us? Uh, really what they were trying to get to is patient-centric, as you described, patient-centric interoperability. Do these rules. I obviously, I think we believe that they're taking us in the right direction, but you know, we, we've seen this before where the, the federal government tries to set a floor, but the industry doesn't react.

Will the industry react to this? Right. Well, so let's look at the c m S rule, right? That's only 129 pages. Much easier to describe. And what it was mostly about was giving us a p i access to payer databases. 'cause although we've done a fair amount of a p I work the Argonaut project on the provider side.

The Da Vinci project on the payer side is, is novel. And so we saw with Blue Button 2.0 some Medicare data flowing, and now in that rule, they're saying, in fact, if you're Medicare, Medicaid, Medicare Advantage chip, any of these government funded or subsidized healthcare programs, you really need to offer this payer based claims a p I.

So will that happen? I think it will. Um, it, the lift technologically is not that high, and of course, with every regulation, this is a notice of proposed rulemaking. You have to see in the final reg what the penalties and benefits are. One would guess, given the flavor of the current administration, it'll be more on the penalty than the benefit.

Right. I don't think there's new budgets for this . Right. So the only concern I have truly about the c m s rule, Is that state Medicaid organizations are traditionally extraordinarily under-resourced from an IT perspective. And we saw this and remember back in the h i e, uh, you know, Obama days when you had a lot of state funding for HIEs and.

You know, the funding ended up going to the, the consulting firms. It didn't necessarily result in operational IT infrastructure at a state government level. So think we just have to watch the state Medicaid side of that. So again, I think C M Ss fine. You know, we'll see hopefully some notion of enforcement or embarrassment or something.

Unless you do it doesn't seem like too bad a deal. There's also the a D T provision, which you must exchange, admit, discharge, transfer data as a patient. Mom goes through the system and again, totally reasonable, technologically easy for providers to do that. Uh, HL seven V two a D T messages are ubiquitous and cheap.

So Alex, c m ss in general. Yeah, so it was interesting to talk to some of the, uh, some of the people in the, uh, innovation community and say, you know, does this impact you? And almost to a person it was like, uh, not really. I mean, we, we sort of had to figure this out, you know, five years ago or we wouldn't be in business today.

So, uh, it, it's helpful and. But it's, it's not gonna dramatically change their, their, their existing business model. Do you think there's gonna be new business models like you and I are gonna get together in a garage that someday they'll take a picture of and say, apple, this garage, and this is where John and Bill started their, you know, their consumer centric, um, application that really transformed healthcare.

Do you think, uh, that this will spur on the creative energy of the innovation community? I do. And here's why. So all of this is just foundational data flow. This does not turn data into wisdom. So where you're gonna see the innovation is when there are what I'll call data stewards that are aggregating this information and then turning it into action.

And let's, let's just give an example of how that might work. You probably saw two weeks ago, Aetna c b s Caremark and Apple announced an alliance. And the idea with that is Aetna will give every member an Apple watch fine. So they'll get data. And c v s Caremark is about a service and a data provider.

Apple will end up hosting de-identified aggregated data and doing machine learning on it to figure out what motivates people to wellness. 'cause maybe, and I'm gonna totally make this up, right, if you're a 30 year old, you wanna coupon and if you're a 40 year old, you don't want public embarrassment. You know your friends finding out you're lazy, who knows, right?

And so imagine a set of services that are offered by companies started in garages that are now aggregating a lot of this new data that we're generating. 'cause patients make it flow and then turn that into action. Better healthcare navigation. I think that lots of startups will start doing that. Yeah.

That's gonna be, that's gonna be exciting. Um, all right, so I, I actually, I'd like to talk a little bit about, uh, a, a different trend, um, than the some that you talked about and one that we were talking about prior to coming on this show. Um, we were, we were sort of talking about, uh, consolidation and the changing of roles.

And one of the things I've heard pretty clearly is that, uh, innovation, digital, and. Uh, information are becoming really three roles within the organizations. 'cause the, the roles are so big, the Chief Information Officer, it's really focused in on the, uh, on the E M R and the data side. The digital officers focus in on the consumer engagement and consumer, uh, experience side, and the innovation officers really focusing in on.

Uh, innovation, uh, really trying to look ahead and to spur a community of people to innovate around the problems that specific health systems, um, are, are challenged with and are addressing, uh, are. So we talked a little bit about this. If, if you're willing to share some of your, you know, your, uh, current direction, I think it speaks to that.

But I, I, I'd like to hear your thoughts of. You know, where you see the different roles sort of evolving, uh, because I heard a lot of this from the floor. I've talked to a bunch of CIOs and they're saying, you know, my role is changing. In fact, your quote from last year, which was our number two quote for the year from the show, was The role of the c i o has changed.

Totally. And, uh, I'd, I'd love, I'd love for you to just talk a little bit about that. Sure. So think about it, in a 19 96, 97 timeframe, when I became a C I O, it was about provisioning, compute, and storage, power and cooling, and then it emerged from there into, oh, security and resilience and disaster recovery.

Then it went into compliance. Well, now it's what? Cloud service for E H R, for email and productivity software for financials. Do I procure? Not provision. It's not CapEx, it's opex. So as you say, you know, back then, 9, 6, 9, 7, I was the C I O, the C M I O, the cso, the C T O, right? And so you went into the early two thousands where the C I O role started to split into multiple people in just that realm.

But now exactly right. We're looking at what is the digital experience for the consumer that's a little bit tangential to keeping the trains running that the C I O might do. And then what is the next business model? Is it telemedicine, Telecare? Is it home monitoring? You know, what is it? And that's innovation.

So as we merge Beth Israel and Lahey, uh, and that merger takes effect on March 1st. I will end up being that innovation person focused on questions like, you know, does Alexa, Siri, Google Home make a difference in say, the workflow of healthcare? If an elder can say, I need an appointment with my cardiologist in two weeks, uh, can you make it in the morning and it just does

Or instead of having people drive to a physical location, you're triaging based on a machine learning approach and figuring out what can be done virtual. Then doing, as Kaiser has half of your visits in a virtual fashion. In fact, we move from building beds to building ISTs, clinicians that are sitting in an office like the one you're in now and doing Teleconsultation.

All of those are untested business models, and they really require a fast failure. You know, let's try it. Let's see what works. And a c I O probably isn't into fast failure. Yeah. So you're gonna, so in that role, I would expect you're gonna spend a lot more time with, uh, innovation partners. Uh, you're gonna be, uh, talking to, you know, different voice.

Providers out there seeing what innovation they're, uh, doing, but you're also gonna be listening pretty closely to the specific, uh, strategies that your new health system is going to be doing and trying to marry those two things because, uh, you're pretty pragmatic. You're not gonna be talking about something that's 10 years out and, you know, waiting for it to come.

You're, you're really gonna be looking at, I would assume innovations sort of layered over, over years, I would imagine. Of course people say What's happening five years from now? It's like, are you kidding me? You know, imagine predicting the internet in 1993, right? I mean, the worldwide web, I should say. So what I look as six quarters ahead, as you say, it's very pragmatic.

It's a technology that exists today, but just isn't evenly distributed, right? So that is, of course, we're seeing ambient listening being used in homes. To connect your thermostats or your cameras, your doorbell. We're just not seeing so much of it in healthcare. So it's not unreasonable to believe that six quarters from now we could have outpatient services running on Alexa or inpatient workflow happening with Siri.

So let's explore those. So as you say, it's the Apples, it's the Googles, it's the Amazons. See what they're up to, how it could apply to healthcare, but as well, These lovely startups run by 27 year olds in their garages that aren't as cynical as you and I. They don't actually know what's impossible, and therefore you gotta watch what they're doing.

I'm in Israel a fair amount too. There's a lot of fascinating Israeli companies around telemedicine, Telecare home care, internet of things and such that we have to explore. Yeah. So you probably went to a lot of meetings that were non-disclosure. I'm gonna ask you for like the, what's the one thing I'll share?

The one thing that I saw that I thought is, is really exciting. I'd love for you to share one thing. The, the one thing I saw was, uh, the nuance booth did a, uh, complete visit without the computer in the, without the keyboard in the room. And you had essentially, you know, a device with the microphones, uh, mounted on the wall in the room with a couple cameras so it could direct the microphones and pick up nonverbal cues.

And it was essentially, uh, catch capturing the entire. Uh, transcript on one screen and on the other screen it was actually using machine learning and AI to, uh, to write the note. And to, you know, essentially put that up there so that when the doctor was done, they could just approve the note and move on to their next visit.

And that's just something we've been, uh, talking about for, for years. That, you know, this experience of looking at the back of the head of your doctor is not gonna cut it moving forward. So voice, uh, gives us a lot of, uh, possibility as you've discussed. Is there something that you saw that, that you were excited about?

So I will take your theme of using natural language processing, ai, machine learning to craft a less burdensome clinical experience to even tell you about a larger trend, which is machine learning, not gonna save us all right? Gotta be very careful about that. You know, uh, wouldn't wanna criticize any particular company's marketing strategy, but the likelihood that Dr.

Watson is gonna read a thousand articles and treat you tomorrow with no human intervention. Isn't happening in the next six quarters , right. What's happening in the next six quarters is I can say, oh, I have studied a million patients like Bill. And what I know is that I can improve Bill's lifestyle if I make these two or three interventions.

If I offer these two or three incentives and, and that kind of thing, the patients of the past informing the care of the patients in the future gets us closer and closer to a personalized medicine approach. We're already deploying a dozen such projects at Beth Israel Deaconess, and it's simple things like, how do I schedule the, or?

Who's gonna show up to the appointment? How long are you going to be in an inpatient setting? And maybe we can schedule all the events in a Gantt chart and not randomness, or how is it that I can figure out for the wellness care that is going to reduce total medical expense and improve quality that I can put you through the right preventative rather than, you know, curative kinds of measures.

All these sorts of things we're doing with existent 11 petabytes. Of patient identified data hosted at Amazon Web Services, Google Cloud, and other places under BAAs to figure out the possible and it works. Yeah, it's interesting. I, I had a couple CIOs, well actually I heard of one major system, health cio say, you know, I still don't trust the cloud.

And I'm like, Wow. I, I just can't believe that we're at this point. Uh, you almost have to trust the cloud for starters, and you have to figure out a way to secure that data and secure the transmission and, uh, the, uh, protect that data. But there's just so many possibilities that are available to you. Uh, the other thing I heard was, you know, people saying, you know, I see, well, Google had a big presence.

Amazon Web Services has a big presence. Presence, uh, you know, the, the Uber and Lyft, you know, so you saw some of these companies start to come in and they had, uh, larger presence. And people are, people were asking me, you know, what's the practical application of, you know, go partnering with Google? I'm like, I.

Well, you know, quite frankly, just go back and watch the show with John and I talking about it. And you shared how with that data you were able to, uh, you know, you were able to look at things and then use fire to push, uh, you know, different things back into the E H R to bring it back into the workflow.

And I think that's, You know, that's the role of the, the c i O now is to say, well, in partnership with the chief data officer and strategy is to say, alright, we need to tap into machine learning and ai. Yes, it's not gonna solve the world, but it's, it's going to be incremental, significant. Uh, advancements for every health system.

So you have to figure out how to tap into it now, be playing with it now, get your wins so that you know when it does, you know, when that curve starts to really take off and you're able to do some pretty significant things. You're, you're right there at the starting line. I, I would think well, and absolutely.

So imagine if the E H R . Instead of being a dumb database, which is kind of what it is, it's transactional systems, gets the bills out, keeps you compliant. You were able to do semantic search and say, you know, a month ago I saw somebody with a gallbladder problem. Um, what happened to them? Now that would be a useful tool.

So when you think about what Google does sort of at their core competency is around structured search. So what if you married Google's technology with E H R technology and suddenly it became a question and answer platform and not a data entry platform. Yeah. Do you feel like you've, you selected a w s, does that sort of lock you in or are you gonna be able to tap into Google as well?

Of course, we selected everyone, , so I have data on a w s I have data on Google Cloud platform, and with the merger with Lhe, presumably we'll end up with data on Azure. And also, uh, all of my community hospitals. I have five community hospitals. All their data is cloud hosted Meditech runs in the cloud, right?

It happens to be n t t data was their original partner. So the answer is, is that you, it's a right problem, right company And I don't have any exclusive arrangements with anyone. Yeah. Alright, so, uh, another trend that I, I sort of picked up on was social determinants of health and health systems. Trying to sort of tackle that.

And you know, we heard, you know, roughly 80% of outcomes are non-medical, family history, personal decisions, financial education, housing, those kinds of things. We've actually talked about this before, but I wanted to take us in a little different direction and let's assume that the financial models are aligned.

I know that's a. As, you know, obvious the business model workflow changes that need to happen and adopting a new business model within the health system are important. But I wanted to take this technology, uh, standpoint, and so I wanted to isolate those variables so that we could talk about, um, You know how a pragmatic c i o would be looking at that.

So I'm assuming we need new data sets and ability to capture data from new sources. Um, we need to be, you know, picking up stuff from monitors, patient reported outcomes, and we need to be able to figure out how to, you know, how to. Really go through that data, crunch that data, and deliver insights to the point of care.

So, uh, since you're a pragmatic guy, how would you look at this as a c i o given that the business model stuff has been taken care of? You know, are you looking at Plat, uh, platform? I assume a platform, and, and where would you go? So you've asked the pasting question. So we have all these new data streams, novel data sources, everything from your Fitbit to your sleep monitor in your mattress.

You know, Nokia makes one of those, but what do I do with it? Right. Do I put it in the e H R? Is it part of the medical legal record? Hmm. No one really knows. Uh, what if I don't act on it? Do I get sued? No one really knows. So our approach for the moment, and I just say for the moment, is to store the raw telemetry from patient generated healthcare data and devices in a separate database.

Then run rules against that separate database, and when we generate an alert or a reminder, then that goes into the medical record. So putting 10,000 normal blood pressures in the medical record probably doesn't help anybody , but the fact that your blood pressure is significantly higher or lower. Now than it used to be.

Ah, well that's something we should act on. And a clinician could in fact avoid information overload or alert fatigue if they just said, oh, I get on average two or three of these a day and they're all actionable. . Yeah. So that's been our approach at the moment. So, so you're interesting in that you, uh, you've actually coded an e H R and uh, Not something you're recommending people go out and do today, but one of the things we hear over and over again is, oh, well, you know, eventually the e h R provider will do this in this kind of, um, let's just call it a, an era of faster movement within health and health IT and health business models.

Um, uh, you know, how, how are you thinking about innovation? Are you gonna be innovating outside the E H R or, uh, I would assume that you almost have to innovate outside the E H R or you're gonna be tied to their, sort of, their timeline of innovation. And you're exactly right. So again, e H R is transactional systems.

Compliance built will exist forever. But do I expect the next great innovation is gonna be done by an Epic or a Cerner? A Meditech eClinical works Athena. Probably no, right. They're going to build their good transactional systems, which we all need. They'll build the fire, c d s hooks or the fire APIs for data gets and puts for which then innovators connect and create value added services that may run in the cloud or on your phone, and that's where the innovation's really gonna happen.

We'll call it the Apple builds the phone, but the app store is where the innovation really happens. It's the app store for health. We'll see accelerate over the next six quarters. Cool. Alright. And John, uh, you know, what, what other trends, what, what other trends do you think we should touch on before we close out the show?

Sure. We can't close out a show without mentioning privacy and security. And the reason being that if we're sharing more data with more people for more purposes, and we're using cloud and we're using mobile, and we're using Internet of things, 'cause IoT actually also means Internet of Targets. Um, in effect, we've created our attack surface.

Uh, you know, 10 are a hundred times bigger than it is today. And so you need to put in place the intrusion detection and prevention, the monitoring, and in fact, probably machine learning driven systems that look for aberrant behavior, not just Port 4 43 has got a lot of data, but the transactional orchestration doesn't seem to be quite right.

Radiology images are being sent to the E K G machine. That wouldn't make a lot of sense , right? And so I see, as I said, I'm in Israel a fair amount. An explosion of cybersecurity companies with novel technologies to help us build a foundation as our attack service increases in size. . So that's gonna be the, the place where we see, I think the, uh, oh, I, I assume the, the, the majority of AI and machine learning, uh, being applied to healthcare in the, uh, as you say the, the 3, 6, 9, 12 month timeframe, um, as opposed to clinical, we're gonna take baby steps, but in security, it's probably gonna be, um, pretty significant I would imagine.

So let's take for example, um, an imaging device, CT or M R I scanner. I'll give you a challenge. Go call up GE or Phillips or Siemens or Fuji or whoever and ask them, so what protocols do you use on what ports and what orchestration can I expect? And you know the answer you're gonna get. Uh, we don't know, right?

So there's no choice but to put in these machine learning services that say, I've looked at the last million transactions out of the cts and the MRIs. I kind of know what to expect. Here's one that's weird. That's the sort of thing you're gonna see this year. Yeah. And, and you also talked about the, the, the, um, the vectors.

I mean the, the vectors are all over the place. I our data is now, uh, I don't wanna say scattered to the wind, but our data, as you mentioned, is, is everywhere. And do you have to sort of build your neural net, if you will, around all those directions that you're sending your data? Yeah. And, and of course there's no way we could figure out ultimately as we go from business associate to business associate to business associate, how data ultimately gets used, but we can certainly control our borders, right?

So that is in my case, you know, with, since with the merger, I'll have 13 hospitals and 450 sites of care. You know, I'll, I'll have to control all the data flows in and out of all those borders. Then put in strong policies, business associate agreements, and auditing of third parties to help me understand what happens beyond the borders.

All right, last question around innovation. So, uh, a lot of different health systems set up innovation arms. Some are vc, they look like VCs, some look like just big sandboxes. You have Techstars, like, uh, Darren's doing in Cedars, um, uh, which is, you know, incubating, uh, small companies. What do you think your innovation arm is gonna look like?

Sure. So my model's a bit different, and that is, is that I am not an accelerator or an incubator per se. What I do is I find third party companies, whether that's a Google or an Amazon or a C V Ss, Or an innovator in Israel and bring their technologies inside the organization where a team of a dozen clinician engineers, these are folks, 'cause it's, hey, it's Boston, right?

Where Harvard, m I t, uh, trained people then dissect those external technologies to look at the art of the possible. So we actually don't even call it innovation. We call it exploration. And that's hopefully gonna tell us what will work in the next six quarters and not, so you're not looking to monetize as much as you're looking to solve the problems that your health system has.

That's it. It's solve the operational business problems inside the organization using IP that may start outside of healthcare or start outside of us. Fantastic. Well, John, as always, thanks for coming on the show. Thank you for coming off the farm and, uh, spending some time with us. I appreciate it. Uh, this show is production of this week in Health It.

For more great content, you can check out the website at this week in health it.com or the YouTube channel. Easiest way to get there is this week in health it.com/video. And, uh, please come back every Friday for more news information and emerging thought with industry influencers. That's all for now.