We're right on time. Thank you for everyone for joining us at this round table. And the topic is managing data overload with AI and wearables. I'm moderating, and that's the easiest job because I have experts surrounding me here. And I'm Kamala, nothing to do with politics, but Kamala from Texas, and I'm an EP physician. And data overload is what we do daily. And just one at a time, if my panel is starting from my right side, if you could introduce yourself. Thank you for having me, Kamala. This is Martha, I'm the nurse practitioner that manages the device clinic at White Plains Hospital. I'm the assistant director of the EP service there. Hey, I'm Arash. You can hear yourself, it's kind of weird. I'm Arash Harzan, I'm a cardiologist here at Emory. I'm not an EP physician. So I don't know what cryoablation or aflablation is, if anyone's gonna ask me. But I also work in the VA nationally. I'm involved with a lot of our national innovation and AI efforts. My name's Hawkins Gay. I'm a cardiac electrophysiologist at Northwestern University in Chicago. But went to Emory for med school, so it's good to be back in Atlanta for the conference. I'm George Stewart-Mendenhall. I'm a cardiac electrophysiologist at Scripps La Jolla, in Scripps Memorial Hospital in La Jolla. I'm also the chief medical officer for Everbeat, which makes it a wearable arrhythmic detector. And I'm a chair of the interoperability group at Heart Rhythm Society, trying to make machines talk to each other and solve the problems that we're talking about today. I'm Niraj Sharma. I'm a local here in Atlanta, Northside Hospital. Cardiac electrophysiologist. Been practicing EP for now 22 years and seen a lot of changes happen in our field. I'm Christopher Chung. A cardiac electrophysiologist based in Toronto, Canada in the University of Toronto. And excited to talk about this topic and also hope to provide perhaps a little bit of an international perspective as well. Excellent. So, I mean, this is a round table, so we can discuss openly. But just to start the dialogue, each of you, some of you are innovators, developers, and some of us are the interface and our patients are the end users in this game. So how do we, starting with the developers, with the data is created, two issues that come in my mind are accuracy, validity, and sustainability of this data. So if any of the innovators, maybe, Stu, I'll ask you to start commenting on this. Yeah, I think as more of these medically adjacent companies become into the sphere, clinicians and physicians and everybody dealing with the data in any sort of way has to become familiar with where the data is created and how it is able to impact clinical care. For instance, the Cardia, one of the earlier devices, wasn't able to judge things like QT, but the Cardia 6L can judge things like QT interval because it has more projections of the lead. Similarly, Apple Watch, it may not be appropriate to judge QT interval from there, but certainly you might be able to see PR intervals from there. So the nuances of handling the data that comes in is unfortunately tied to the way that it is created. And as time goes on, I think we'll get more and more sophisticated on what actions we can take from the data and the knowledge of where the data comes from will increase. Unfortunately, we're still in a little bit of a wild west time in this with FDA cleared and even non-FDA cleared devices that are able to provide pretty good data, but we have to know where it's coming from and what it's able to do. And maybe I'll make a comment. I think this is a really interesting area, but I thought there's a lot of gray area here because a lot of these devices were developed for the consumer in mind, right? It's a consumer oriented device and certain aspects may not be FDA cleared, but at the same time, we're seeing patients use them for clinical purposes, right? And that's where it gets a lot much more unclear because how accurate and how valid are these devices when we use them in the clinical sphere when they were actually designed for the consumer. And there are also just individual levels of comfortability between physicians. So I know some of my colleagues feel very comfortable making decisions off of patients who send them Apple Watch tracings, but then there are other physicians who know more about the Google Fitbit and they're like, I'm much more familiar with the algorithms on that and I don't know as much about Apple. So right now there's individual clinician level comfortability with like consumer generated health and what you'll trust and what you won't trust. And so it's a little all over the place. If I may speak about the impact that that has, thank you for saying that because I have entrepreneurs actually coming to me and asking me what do I think about the app that they're creating? One of them actually looks at the QT interval for the express purpose of loading tickets in, so all of those things are great from the perspective of being at HRX because we are now being asked and we, I like to think of ourselves, I am an AHP. So we are the boots on the ground for a lot of this because the impact, it's not only on the end user, the patient, but also it's on the clinicians. The patients will have a quicker access to your nurses, to your nurse practitioners, to your techs and they will to the physician who, you know, my guys are in the lab. So I think that the impact that all of this technology and I think we're in a transitional period, it's very telling. We need to report how that impacts us because we are the, not just the interface, but we are the human interface to what is happening and our patients absolutely deserve more education than Tim Cook is providing for them when he's selling them that Apple Watch. Go ahead. Good points there, Martha. You know, going back to the original intent, which was pointed out that this is for consumers, but I think what's happening now is the data that is provided from any source, take it from an Android or from an Apple, the initial intent may have been for something else, a fib, but now what's happening is we have innovators that are taking the data that was meant for maybe a fib and using that data for other purposes, like you mentioned for QT and table, but it goes kind of beyond that for even for PVCs. These algorithms obviously from Apple were not designed for that. So now we have innovators and individual companies taking the data and taking it to another level. So I can see this happening in the near future where data is freight from Apple or Samsung into another party, third source app, where they take that data and then they, you know, take it from there to do, you know, other algorithms that they are probably developing. Can I just add something, Kamala? I work in pediatrics. Absolutely. So we take a lot of transmissions from our patients from their Apple Watch. Kids are not approved, not FDA approved, so they always have to lie about their birth date in the app in order to transmit to us. So we are taking the product that's out there and using them on children and I agree that physicians are very comfortable with taking a transmission and taking a kid to the lab because we diagnose SVT and more and more physicians are getting more comfortable, but this is the way we catch rhythm problems in kids. They don't have a high burden of AF, they have SVT here and there. And so hopefully the evolution of these products will continue and, you know, we'll be able to use them more broad range. Little babies have Owlets, which is similar type of thing, but a sock, but the kind of teenagers all use their Apple Watches or even their parents' Apple Watches to give us the data that we need. So hopefully we'll be able to expand a little bit more for different populations and not just be a fib, you know, moving forward. Thank you for all the comments. Arash, do you have anything to add to that? You know, from the regulatory standpoint, any other comments to just the developer side? Well, I can't speak to the regulatory side of the house because Novia doesn't really have any role in that. But I think from the, you know, I think there's this concept of drip. So your data rich and information poor. And that's something that, at least in the VA, ironically, we deal with a lot, where we have, you know, probably, you know, more multidimensional data on our patients than anybody else has on their patients. But, you know, harnessing it is really a challenge for a variety of, you know, VA specific, but also I think just, you know, data infrastructure reasons. And then this is now sort of really amplifying in this, you know, brave new era of not just patient generated data, but just, you know, we're finding new data sources to make use of. I think it's almost like people just realized that like, say your, you know, ICD collects, you know, a lot of impedance and, you know, I took your data, right? And we're trying to find the uses for that. And so now we're basically trying to pull in, you know, more data into what is already a, you know, disconnected system. And we can't even, you know, you can't really make sense of your, you know, very relatively straightforward EHR data. Throwing in like new data streams on top of that isn't going to make things easier. It's just going to make things worse. And I think there's, from the developer side, I think what, you know, really I think frustrates us is, you know, everybody wants to be the platform of the thing for the future, which is a very admirable goal. But I can tell you, like, you know, the VA, when I'm involved in these conversations, we don't want to buy someone's platform. That's something that, you know, because you can't buy everyone's platform. You have to, at some level, have an ability to interoperate. And I think that's one thing I think developers might, you know, think about is, you know, how do you really, you know, try and think a little bit outside of the traditional mold of, you know, the platform play is probably a very like smart business one, but that means everyone's got to be on your platform. But if you're going to, if you're, if the person you're selling to has already bought someone else's platform, you're excluding yourself from that market because they're not going to want to now buy a whole, they just want to be able to pick and choose what works best for them and just plug it in where it goes, right? So I think that's the biggest developer facing frustration that at least I've had. And this is true all the way from the startup to, you know, conversations with Apple, right? So everyone's got the ecosystem and the platform. So congratulations. We can't buy everyone's ecosystem and everyone's platform. And just to continue on that conversation, I think there's a lot that we should advocate for when we work with developers to try and make sure that we have a high quality data that is recorded. And we get into a little bit in our perspectives article is, you know, there's so many ways to collect data. And, you know, the obvious example is, you know, all these PPG monitors that you can get, but we really should be advocating for as high quality and granular data as possible and wherever possible for a rhythm, you should be getting ECG data, right? You should not be making as many conclusions on PPG data because that's where we open ourselves to false positives and then all the associated challenges and, you know, complications that could be resulting from that. So we should, as a community, be advocating for high quality data and developers and regular citizenry should be making ensuring that as well. If I may add something to that, I'm sorry. You know, just what Christopher was saying, I think that that's something that we as clinicians can demand from startups and, you know, innovators. But, you know, the other thing that I think nurses have a clear view to is the impact on their patient's psyche. You know, we are doing all of this for our patients and our patient care, and we want to keep patients healthy out of hospitals. You know, talk about patient population health and all of that. But, you know, all of the creations that are going on here, as spectacular as they are, and they are, are creating an impact on that patient's psyche. You know, we need to do better by the patient. So to me, everything that's created should just be a foregone conclusion that it's going to be completely centric to the clinician. Because I think that we're missing the bigger picture, the bigger boat is that patient has just so many different expectations when they buy whatever the apple of the day is. Absolutely, it's a great point. So, you know, towards the end, we actually had, you know, I had these ideas of gaps and generational gaps or even psychological impact. But I think Stu wanted to touch on. I think that's a very important point. Let's return to that in a second. Just, I wanted to comment on something that Neeraj said at George Mendenhall about the abstraction of data. I mean, should these all be dumb pipes that an EKG is an EKG that goes to some central processing to allow, you know, vendor neutrality for the VA to purchase an EKG, a wearable EKG is a wearable EKG. I think that ultimately where, maybe where things go. Unfortunately, as we said at the beginning, every device is so different in their resolution, in their ability to pick up nuances of EKG, in their ability to determine QT intervals or even just a rhythm versus not rhythm. That we're not at that point yet. And first generation cardiac is very different from an Apple Watch in the resolution and the way you could trust a P-Wave from it. And that's mainly because of technical limitations. So even without the technical limitations, a arbitrarily placed chest strap may be approximating V3, but in some person it may approximate V6. So without really sophisticated, knowledgeable machine classifiers that are knowing the source of the data, it's not gonna be possible for the average clinician to know where the data is and what they can and can't tell without kind of more binned guidelines that are tailored, unfortunately, to each data source. But that kind of already happens with technology that we use already. So like, I might feel more comfortable diagnosing something off of a, you know, an iRhythm Xeo than I do off of the like, Bardi Cam Patch or something like, I'm just bringing that up. Like, everyone has different levels of familiarity with different products that are out there. And you just have to get the data. And then when you're looking at it yourself, you can make the decision what you trust or don't trust. But I do think we will also eventually be building algorithms that aren't necessarily part of, you know, Apple has some of its own algorithms that are on its watch that can make some diagnoses for us. But if there's an efficient manner for us to store data that's sent to us on a central platform at the hospital, there can also be algorithms on that platform that can utilize that in different ways, you know, that's based specifically on the hard data that's stored in the platform. That's probably something we'll touch on. Like, how do we store all this data eventually? But, you know, having different data sources that, you know, people feel, you know, more or less comfortable with is something we already have to deal with. Yeah, I mean, we talked about, you know, some problems and some dialogue around it. Just in your experience, anything, any comments on EHR integration? Because I'm thinking of the clinician burnout or overload, data overload from the clinician standpoint. Any technology over there? And just comment based upon your experience and I can start with you, Christopher. Thanks. I mean, I think this is an area that we still certainly need to work on. I think there has been a lot of guidance from the devices, from the device space, especially from cardiac, electronic, implantable devices. We see a lot of developments in remote monitoring. And so there's good pathways there in place to manage high volumes of data. Now, obviously, wearables introduces an even higher volume of data potentially. And so we need to work on developing those pathways, right? Unfortunately, part of the data silos that emerged are because the generalized EHR is not able to categorize every single type of data, right? That would be a Herculean task in the 1990s and it's still very difficult in the 2020s, right? So when most of us go to Epic or Cerner or our main EFR, there's a different PAX, right? So you go off to the PAX server and you see the chest x-rays, or you go to Intelespace or whatever you're using and you see the echoes. And the main reason that's fragmented like that is because PAX is better PAX than Epic would be displaying PAX in clicking on something and waiting for it to appear. It's probably loading off the PAX server. So part of that is a technical thing that as things become more integratable and it's more transparent to the user to click on an Epic message and go and see the echoes and see all the echoes versus dates. But that is a model that might even change. We might wanna see snapshots of the patient and see their EKG and their echo and their wearables over time. We might wanna scroll through this and see their different, their QT evolution versus time or their other heart failure indices that become aware. So this is partly technically limited. And I think that the fragmentation and the silos are mainly because it is easier to log into the Geneva, to the merge, and see all your devices at that point rather than go across back and forth. Now, we saw some of the startups that did have some putative solutions for that where they will bring in contextual data when you're looking at the interrogation for the remote interrogation for the device. And that will certainly help most people's practical day-to-day lives or the workflow and help with burnout and things like that as time goes. So it's a partly technical, partly social, and partly waiting for integration to happen. But this is something that we have to guide as clinicians and researchers over. So can I ask a naive question then? Oh, I'm sorry. No, just gonna make a comment. This is more than likely what's going to happen. You would have data from different sources, we'll just talk about Apple, you'll have EKGs coming from there and they would go into EPIC, CERN, or whichever EMR you have and there will be a module. There would be a module that would be designed to look at these EGMs and it'll spit out a diagnosis which would be much more advanced than what an Apple Watch does. So this module you would have to purchase separately, it will be an add-on to your EMR and it'll spit out something and you would have to over-read them just like you over-read EKGs and you agree or you don't agree or take it from there. But that's what I foresee, well I know this is what's going to happen, that's where we're headed towards with this huge data load. And one more point to that, we kind of briefly touched on that, is these algorithms for Apple Watches and Samsungs and others, Fitbit, etc., they came out not really informing the patient about its implications. So our patients, we'll talk about the psychological impact later on, but one of the other impacts is how to use it wisely, right? So we have to educate our patients, do not send all these tracings and unless there is education and unless there is some sort of a, I don't want to say a carrot-or-stick approach where you can't be abusing the system and sending data constantly unless there's a cost factor associated with it, right? So if you send a transmission and you think it's real, go for it, but there would be a cost attached to it, I don't know, to be worked out, right? CMS, etc., probably will be slow to act. You know, we all have in our office certain rules that we place, right, you know, if you come late or you cancel your appointment, there'll be a charge, etc. If you send a transmission over, there'll be a certain charge. But the point I'm trying to make here is that there will be a mojo, there should probably be a charge or some sort of patient responsibility when you do send a transmission over. I agree with that. I was going to ask a naive question, but I think you answered my question in that there isn't a universal programmer, there isn't a universal EMR, there isn't, I don't think that that will ever come to pass. It will be, like you just said, modular and it's just patient education and cost. I wanted to ask, too, being part of the interoperability group with you, I don't get to ask that many, too many questions when that group is going on because, frankly, I'm intimidated by early brainiacs at that group, but how is that going to be resolved if, you know, all these apps may be speaking a different language? Because that's what we're working on, right, in the interoperability of all these machines talking. Yeah, I mean, I think that's the core of the problem. I mean, to some degree, certain things are very clear, right? When you get a potassium value, you don't care if it was run by LabQuest, Quest, or, you know, your institution or wherever, right? Potassium is potassium. It's been normalized, it's been studied, similar to the evolution of the INR when Coumadin was being monitored, right? So similarly, we will start to get very scrubbed canonical data. We'll know that this is a clinical-grade ECG, it was a rhythm-determined EKG, it doesn't matter if it came from an Apple Watch or a Samsung or, you know, whatever du jour, and we'll be able to do that. So the hope is that the semantic information is preserved, but the nuances of where it was obtained or how it was obtained becomes more abstracted over time for the utility of the clinician. Thank you. I was just going to say that I agree that all of this data is going to be stored on, whether you call it modules or dashboards, or there are companies who are working on creating these dashboards that can incorporate data from all of these different devices. And there are frameworks that have been created for interoperability, like the Trusted Exchange Framework and Common Agreement and the Fast Health Interoperability Resources that protocolize, like how you want, how data should be shared between different trusted health information exchanges. And if those kind of structures and protocols are followed for creating this data and those dashboards utilize that, then they can be shared between dashboards. And, you know, like Northwestern might pick to partner with one dashboard and, you know, Mayo Clinic chooses a different dashboard. But if it's structured in the same way, it can be easily shared. And, you know, Epic is already starting to give users the ability to download their own data onto apps because of those protocols that have been created to allow, you know, information to be exchanged more easily. So we are starting to develop protocols for storing and structuring data in a way that it can be shared. And I think the dashboards that are created will utilize that, almost like, you know, like Apple HealthKit. Like if you want to build an app that is going to be on iOS that can be stored in the Apple Health app, you have to follow certain rules that Apple has created so that your app can exchange data with the health app. And that's the point of the, you know, the Trusted Exchange Framework is that you have to follow these certain rules. And if you do it, then the app that you build can share information with, you know, the dashboard that the hospital is using. Harash, I think you had a comment. Well, yeah, I was just, while Hawkins is talking, I was just thinking about, like, the, like, this sounds like a movie called, like, Dashboard Wars, where, you know, and it's kind of what we have, right? But, you know, the, I think going back to, I guess I'll be the guy that just keeps harking on, you know, industry, because I'm the government guy, and that's what we do. But, you know, like, take Pax, for an example. So at Emory, like, we've got Fuji Pax, for better or for worse, with, you know, GE and Phyllis machines, right? But all the images flow. And the reason for that is because there's a standard, right? Like DICOM exists for this purpose. And that's something that industry came together and said, here's what we're going to create as a standard. And now I think the wearable, you know, AI, you know, data space is obviously very new. But I think that's what ultimately has to happen is that there's got to be a certain level of commoditization. So I think a lot of these are going to become, or are going to need to become a commodity so they can interop a little bit. And you need to have some standards that industry has to come together and create. Because otherwise, you know, we're going to be left with the opposite, which is, you know, one tripping over each other to try to sell you the thing, and their thing is better. We actually had an interesting conversation, you know, this is National VA, with one, I'll just say big like healthcare IT company, who had built, and I didn't even see it in demo was kind of a conversation, but they had built a platform. This is trying to address the, you know, in the kind of computer vision, diagnostic radiology, AI space, there are so many companies are playing in this space for different modalities of different purposes. So they had created like a way to basically, you know, pull in, you know, x number of, you know, other companies like things to look at, like the modality that you're using, and you can just, you know, upload it into like, this company's like, you know, AI in the cloud to like, let it run without having to pre integrate and buy all these other companies like little solutions. And you know, like, that's, it was interesting. I don't know if it's going to actually take shape. It was a one off conversation. But you know, it's an example of at least like, you know, industry sort of maybe taking an approach of, you know, how do we get our own solutions to play well together. And I think that's what we need more of, within reason, I think we recognize that there's obviously like a lot of, you know, IP and effort that goes in, we don't want to want to feel like they're giving things away, but there has to be a standard, I think standards have to be led by industry, because otherwise, if you let the government get involved and make standards, well, that's not always, you know, what leads to like what we want. I do want to just make a comment about like, you know, we said that, you know, clinicians need to be the ones demanding things from companies. And I don't actually think that really, usually works. Because if that was a thing, then you know, prior authorization wouldn't exist. And you know, clearly, that hasn't happened. So I think just clinicians, you know, wanting to will something, you know, doesn't work. And a lot of that's on us, because, you know, like, we don't really all kind of get together in one voice and speak. So I think we really need to push industry to, you know, get together and think about how do they make it, you know, easier for the rest of us to work with them, because we want to, but we just, you know, can't do it in a vacuum. I just wanted to take a moment to focus on an earlier point. And I completely agree that we need to have standards across the industry to make sure that the same data is being collected and transferred. You know, when you talk about commoditization, and we talked earlier about, you know, does the patient have to pay for this? I think that's a question to discuss, like, who bears the responsibility for this? And how much does the patient have to pay for this? There was an interesting article that was published in the Journal of American Heart Association recently that looked at the psychological impacts of the wearable devices and how patients that use wearable devices seem to have a higher healthcare utilization, more emergency department visits. And so there's a couple perspectives. Is that because the wearable device is driving that? Or is it because this is a huge need that is out there? We know that, you know, 20% of patients may have a hospitalization or an ED visit while waiting for an ablation, even. So there is a need out there, and are these devices filling the need? So how much is it the patient's responsibility to pay for this service? Or is it more that we have to build a system that can take that? That's an excellent point, you know. Yes. I want to just go back quickly to the standardization issue, and you talked about industry coming together. And if you're not familiar with it, in the home automation space, which is kind of a debacle if you've tried to automate your home in the last 10 years, there was 20 apps and everybody talked differently. This is a great example, by the way. They came together through an organization called Matter, and it was involving all of the large home automation companies and created a standardization protocol for how to interact with those devices. And I think you're absolutely right, the health systems, probably as much as anyone, needs to really drive that because it would facilitate the adoption of these tools, which I feel will be of benefit to patients. But right now, if you have to implement remote monitoring system one, two, and three, you walk down to your IT department, and they want to just kick you out. Yeah, well, a quick comment on that. We encounter situations like this when we, you know, through our pacemakers and ICDs, we have different remote monitoring situations, and they all feed into one big, depending on what system you have, it feeds into that system and it feeds into us, right? But the data is kind of uniform, it's proprietary in some cases, and for example, Medtronic and Boston Scientific, they're different, they don't talk to each other, but ultimately we have a system that does give us the information, albeit separately. You know, I think going back to your point, there will be separate modules, you know, different EMRs are not going to talk to each other in that modular aspect. The data that's going to be supplied from the source, your watch, may or may not be similar, may or may not be standardized. If it is standardized, that's great, everybody would have a uniform platform, but if it's not, then these modules would have to kind of integrate both, just like what we have now. We have one source where all the data from pacemakers, ICDs, different manufacturers goes into one source and it subsequently comes out. The same concept probably would apply if there is no standardization, where standardization occurs at the end source, not at the beginning. Mendenhall, so the standards and the incentives are not necessarily completely aligned, but they're getting that way. And looking at history, I shudder a little bit to say that government should never be involved. Any strong opinion either way is probably wrong, right? So looking at history, showing my age a little bit, what we had, you know, Prodigy, CompuServe, and AOL, and they didn't talk to each other, right? And then it was in everybody's business interest to come together. We now, if you go on the web with a Mac, you go on it with a PC, you go on the same internet, and that hasn't been that long. And there were structured and very good interests to try and keep these separate at the beginning, but people realize talking together grows the industry. It makes everybody work more efficiently. And if you stymie that, or you're not at the table, you're going to be left behind when everybody else coalesces to standards. Now, government wasn't dictating all of the nuances of it, but, you know, everybody picks on Al Gore, but the Gore bill did fund a lot of the transition from ARPANET to, you know, BitNet, ARPANET to create the modern internet through funding and advocacy. So I agree that the government should never start prescriptively trying to mandate text fields and things like that. It's probably not something that's going to happen, but having incentives and having overall structures for everybody to talk together for growing all of this economy and growing all of the spheres and opportunities that we have when we do talk together is something that I feel quite passionately about, and it's likely to happen. And again, like the Fast Healthcare Interoperability Resources is a set of kind of structure, like data storage and structural kind of recommendations for sharing data between apps. And if you want to be a participating, like, health information network on the TEFCA system, you have to follow those standards, and that's why they were built. FHIR being a subset of HL7, which is Health Weather 7 Interoperability Group. That's a very good analogy provided there. So just to switch gears here, we all talk about CMS reimbursements and clinician burnout tied to those reimbursements. Now when we all get wearable data, just want to know how do you all handle, because that is an overload in my clinic. So we can start with Arash, and then. I mean, I may not be the best example, because a lot of the work that I do clinically is in the VA, and we struggle with this for other reasons. But I think to your bigger point, we all spend a lot of sort of unbillable time reviewing things, and so how does that work, right? And one thing that we do have in the VA is some sort of quasi-unique things, like we do a lot of what are called e-consults. So instead of us having to go and see a patient, you can actually just request an opinion from a specialty, and then that opinion, someone reviews the chart, puts it in, puts in their couple-of-line recommendation. And it's not, in the VA, we talk about workload. So the workload is captured in sort of like VA parlance. But that's one way that we actually, it's an effort that gets at least seen by the system as a thing that you did, right? And so are there other ways of kind of doing that in the sort of broader pay-for-performance, pay-for-service, fee-for-service environment? Maybe. And maybe that needs CMS to either create some new codes to let people sort of capture some of their effort for some of these things. And I think it's a need, because on the one hand, you can say, well, it's just people being greedy. On the other hand, I think it's you want to be able to incentivize people to spend the time to actually review a lot of this data that's going to come in more and more, right? So you have to balance it out. So I think that's what we need in this case, is a bit more either creativity with the codes that we do have, and I think some may apply, versus like having a hard talk about do we need to start developing new codes that can actually reflect some of the work being done when you're reviewing someone's Apple Watch tracing, right? Then you're reviewing a thing that came from a device that was not prescribed and in itself is not covered by CMS, right? So does that get you in this very interesting space? I don't know. But there's definitely a need for that. Well, I think that's a great question, and going back to your question, how do we deal with this data overload? And so my pat answer is very badly, because there isn't any integration as there is with remote monitoring. So we've had makeshift frameworks that we had to come up with in order to not only look at the data, but now that you're looking at the data, if it's actionable or not, you still need that clinician to communicate with the patient, because the patient, I don't know about anybody else's population here, but my population, it's elderly, over 68. I work in a very underserved community. I work in the Bronx. So it's underinsured, undereducated, underserved. And even if we wanted to create a pay per transmission sort of work model, they just have to choose between the rent and paying this co-payment. So we are, as a nurse, as a nurse practitioner, my guiding light is always my patient care and however I can deliver that. But it is extremely difficult to have a patient send a transmission, because now they are not, if they use a Cardea, they are not with the company. They haven't paid the fee. So we provide them our email address. It's a generic office, generic email address. And so the challenge comes with the education of the patient saying, you can send me up to three or five tracings, as opposed to the patient who can pay for that transmission, who will send me a hundred, but refuse to pay. So it's very, very difficult. We do the best that we can in that makeshift workflow, work frame that we have, and just hoping that there will be other things that come our way in order to integrate it. Like somebody presented an app to me called, oh my gosh, I forgot. It's sort of like a third party vendor for digital data. And so there are creative answers out there. And the point that I want to make here is that they don't have to impact the patient financially from the get go, or the patient will not accept that. They'll just find a way. I have patients walking in with tracing saying, I downloaded this tracing, can you have Martha see it? So we just have to be creative as to how we are going to help our patients. And I think that at this meeting, that's what we see. And I completely agree with that. We need to find a way to solve this problem because that data overload is coming and we need to find a solution for it. And I don't think the solution is going to be charging for each individual strip that is sent. So it's certainly a huge challenge that we need to work on. And I think we can leverage a lot, like we said earlier, that we've done in the CID space, right? And perhaps the AFib clinic of the future will have a remote monitoring arm where patients will be transmitting regularly. And so that certainly has to be built in. Sorry, there is a, I don't know, I'm familiar with one third-party app where it takes your watch strips and it's actually reviewed by somebody, it says an expert person, and then they give their information to the patient. But I'm sure all of us have been in a situation where the patient brings their phone and you're scrolling through their whole log looking at inconclusive or sinus tag, sinus rhythm, AFib to find one little casing, which is probably an artifact showing AFib. And it's just so time consuming and frustrating. And you're right, our patients are elderly and they don't really know how to use the app properly. But I really, I have no solution here. There's got to be a mechanism where we try to tell our patients, we educate them, hey, listen, we don't have time to go through all these strips. Please print out the pertinent strip, print them out. Don't bring them on your phone, print them out and show it to us. We do it every time to try and reinforce that this is the way we would like to do it. And some of the patients have caught on and some haven't. But just to continue off that comment about, and Suhr can talk more about this, but we can have some form of closed loop workflow that actually will, you know, filter out all the, at least maybe all the noise or all the easily actionable data. There's certainly a lot of data that will fall in the gray area where, you know, you do need a clinician oversight, but there's going to be a lot of data that can be potentially very easily actionable that obviously the sinus rhythm, you know, all their algorithms say sinus rhythm and we can reassure the patient, or this is clearly noise and there's no point in even trying to interpret this. So some of that can be filtered by algorithms that, you know, can feed that back right away and can be a strategy to reduce the workload on the clinician. George, I agree with both of what you said. I mean, a lot of it, I've been astounded, and maybe I shouldn't be astounded, but that most, almost all patients respond to education. I mean, people love to hear, hey, don't worry about it, right? I mean, then so they need to know that in a small, if you're at low risk, you have a bleep of atrial fibrillation, it's not 911. And I agree that the manufacturers of all devices, for better or worse, have not done a great job of putting these devices in a medical context. I mean, they want to sell devices, right? I mean, they have an incentive that's not directly aligned with ours, but, you know, we can align it. We can work with that. Education, then automation. Automation is a little bit of a double-edged sword that if you have everything is normal, everything is normal and very high sensitivity, then when the one thing is not normal, then now all of a sudden that's one in a thousand events instead of a 2% they were seeing, and they really don't want to keep that, treat that as noise or ignore it. So, but it can help with that. I think just really education and the informing of the patient and their family members of the context of these and how it is important, but it's the same way as a symptom. You may feel a symptom and it goes away. You talk to the patient about it with your doctor in three months. You don't have to call them right now and send them a message right now and, you know, get everybody worked up. So. Any other comments, Hawkins? Yeah. I mean, I guess that was the whole point of our like perspective article that we wrote for this is how do we, you know, right now we're handling all of this digital health data, you know, in individual ways at individual health systems and it's overloading us and we're not doing it well, you know, in the ideal world, you know, we have these dashboards or modules that are incorporating that data. And then, you know, in the background of that module, there's some, you know, multimodal context filtering AI that can say, oh, someone just uploaded in ECG to the system and this is atrial fibrillation, you know, but this is a, you know, a 45 year old with no other medical problems. And so I'm just going to send an alert in the background to the healthcare team that, you know, someone needs to assess this at some point, you know, versus, you know, in another context, it'll say, oh, well, this is a, you know, a 72 year old who's had two strokes and now has atrial fibrillation and this needs to be acted on more urgently. And so that kind of context filtering, filtering through the kind of urgency of how fast we need to respond to that, it gets, it gets, you know, internalized into the module, there's a first pass at the data, and then the teams, whether it's a remote monitoring team that you've built in your clinic or the physician is alerted with some urgency based on the context of the data that was uploaded. And, you know, these aren't like in the distant future now, you know, like there are multimodal contextual AIs that are being built that, you know, already have a lot of, you know, data behind them showing that they can understand, you know, medical histories and they can filter through unstructured data like medical notes and understand some context about the patient and make these types of decisions. So the infrastructure is not there. You know, we are not storing the data in an efficient way to deploy an AI that can do that. But once we, you know, build the infrastructure and have these modules, you know, having AI work in the background like that, I think it's not going to be, it's going to be a reality at some point in the future. Great discussion. You know, for our listeners, Dr. Chung, Dr. Stuart George Mendenhall, and Dr. Hawkins Gay actually have a wonderful paper with a floor diagram that practically talks about everything we've been discussing today. So it's been a great resource for me. And just one, I know time is running. We have such fun discussion, but just last thing, all the experts here, comments on generational gap and the illiteracy and the psychological impact. They're all tied to, you know, the wearables. We talked briefly about psychological impact. Sometimes, just like Neeraj, you mentioned, a patient comes in with data on their iPhone or, you know, they show us everything. They're so anxious and our time is limited. How do we incorporate that calming mechanism beyond the technological interventions we talked about as a clinician sitting with the patient? You know, do we need a psychologist sitting there or these companies? Because we are EP positions, our time is limited. So if you can just comment. I think it's very difficult right now and it will, it's certainly going to get better because it has to get better, right? I mean, I get sometimes epic inboxes saying, hey, I feel dizzy. What should I do? I mean, I, you know, I sit down with a patient and say, well, is that a question that you think a doctor could answer without at least a phone call? I mean, it's not something that's an appropriate for email. And they're always like, yeah, okay, I see that. Well, I was feeling dizzy. But so, and then they're like, so then they realize that, you know, look, you have to wait. If it's really, really bad, you go to the ER. The ER is always there. It's our backup safety net for everything else. There's messaging, but don't respect real time response. And then everything else you can give us a call during the day. And at the end of the day, I returned my calls. And when people understand that, I think that helps better. So it's a little bit social and then the technological will help. I don't foresee the omniscient chat bot really being that great. I mean, that's a glorified Google search to some degree. And you don't necessarily want to trust it with all the advice that it would give you. So I, happy people are going down that route, but I don't foresee that being the total solution. But then a combination of all the healthcare teams, us and technology and education and social factors maturing, I think will lead to the solution of this problem. Any comments? We can go in a circle or, you know, Arash, you want to comment? Yeah. I mean, I tell a lot of people that since COVID, I feel like the average age of the new patient that I see has gone down by like 10 years. And I've definitely earned like a secondary degree in psychology because, you know, I see more people with, you know, chest pain, palpitations, what we're all seeing. And it's, you know, the worried well. And it's really obviously, you know, I think, you know, everyone has a legitimate concern. I think I've had to look at my job more as just giving people reassurance when I can, because the peace of mind is an important thing. But, you know, we have to recognize at some level that, you know, healthcare is a finite resource. And so in the era where I think we all expect to have like immediate access to everybody. And now I think we're also like, you know, in a consumer environment expect like patients are expecting to have immediate access to their cardiologist, their, you know, electrophysiologist. It's hard to be overly critical. I think when someone's worried, they want to talk to a doctor. By the same time, I think we have to think bigger. And how do we sort of set up a way that you know, we give people the reassurance they need. And, you know, but also, I think, maintain the resources and the and the infrastructure that we have and use it effectively. Because for every 18 year old that I'm seeing with palpitations, that's, you know, one patient with ischemic heart disease that I'm not seeing at the same time. And right, both need to have the care that they need addressed, right? It's a matter of like, who does it? How do we do it? But yeah, I don't, I don't think I have a solution. I remember the time where you had to pick up the phone and call someone and leave a voicemail. And if they, you know, called you back in a day, you know, you'd be pretty happy about it. But I think we've sort of gone to a place where like, there's just a, you know, I think we're all guilty of this as well. You expect like an immediate thing. Right. And so when you're stressed, and you want to get an answer, and if they don't get a hold of the doctor, they're going to Google and that's going to, you know, lead to a whole different set of problems. So I think it's a, it's a huge issue. I don't, I wouldn't pretend to know what the solution is. But I agree. It's a, it's a real, it's a real thing. And it's a challenge. I think we have to, you know, despite that a lot of these may be what we call quote worried. Well, I mean, there are still people out there that have true clinical disease and, you know, their diagnosis may not have been possible without, without the availability of a wearable device. So we have to be able to differentiate and identify all those individuals. And like we said, even those that have what we think is, you know, low grade disease, like a lot of them may still have adverse outcomes. Some of them may have hospitalizations or ED visits, and that has its own tax on the healthcare system as well. So, you know, I hesitate to say worried well, because all these individuals still have some form of disease. Also, I want to speak on an earlier point, you know, talking with generational gap, I think it's also important to talk about, you know, socioeconomic disparities, you know, in terms of, you know, these devices, we need to, you know, there's certainly maybe some demographics may go for more cheaper devices or, you know, lower quality data devices. And that's, that's an important challenge in our, in our system. So we need to, it's not just the generational gap, but potentially, you know, socioeconomics may play a role into what types of devices these individuals are using. And so we have to find a way to, you know, evolve and adapt to that. I agree a hundred percent. That's a fantastic question. You know, I'm just like the rest of the panel here. I, I agree there aren't any solutions that we have found yet that we can put in place immediately to solve the problem, because I think that, you know, COVID was a global education for us all. And the acceleration of everything we needed to do just catapulted all of these issues of accessibility that the patients are now under the belief with how telemedicine came to be from COVID, that they have unlimited access to their clinicians. And I think it's, it's not their job. I think it's our job to just sort of set limitations on, you know, this is the accessibility that happens. And we don't need to speak about finances, just simply educating our patients that we are not infinite. You know, we are not infinite. And that Google, that in basket message that you, you spoke about is exactly what we come up to every day for, for us nurses, nurse practitioners. I think the bigger challenge is that those patients will always have access to their nurses before they get to their physicians, because we are that layer. So, you know, I'm working fast and hard on how to just get ideas from all my, my brother, my AHPs as to how they do this workflow, because it is just about the same song from every AHP that I know. So I think we should just find solutions all together at the table because it's, it's a team, it's a team effort. I agree. Blaming the patient is not something that we should or ever do. And if, you know, if somebody puts a buffet outside your house and says free breakfast, then you start going there every day for breakfast. So then they're like, oh, well, we only met this breakfast for when you're really, really hungry. Right. I mean, so we gave them the access on the inbox. There's no, there's no warnings about it. So what did you expect? Right. What did you expect? So I found that every, almost every single patient responds to the education. I let them know a little bit behind the scenes on how it works and it goes to my, you know, my assistants first, and then they will forward it to me. But if you do it twice a day, there's no way we can continue that service. Every, almost near a hundred percent respond to that. And it's, again, it's, it's, that was not their fault. And I think we need to really address that through education. I'm going to, I'm going to just sort of lend a little more to that thought. You know, I, I do hold administrators in our healthcare systems responsible for, they should not advocate unlimited access to clinicians. They should have, listen, Twitter has 150 characters. So do not allow my patient to write 300 pages as to when they were sick at age five and now we are here. And by the way, I'm having palpitations. I had to read 60 pages before I got to the problem. So our administrators owe us having some limitations into the system that will allow us to answer more messages rather than judge us on, it took you three days to answer this message for this patient. So I want to go over what Christopher said. I think that's, that's really important. When we talked about socioeconomics, these watches are really expensive and for, for a lot of my patients beyond their reach to get a watch. And there are non-FDA approved products out there. You can go on Teemu, right? For those of us on Teemu, you can buy a watch for $35 that gives an EKG, not FDA approved. I have used it because my patients brought one for me to use and asked me if this was accurate or not. And I use my Samsung watch on a patient who had known AFib, put on one hand Samsung and the other one on a knockoff Teemu $35 watch, both diagnosed AFib. But of course, off-label one patient doesn't mean anything, right? But the point I'm trying to make is are we, because of the cost issue, right? Are we doing a disservice to our patients that just don't have the resources to be at par with other patients? That's another thing. I mean, should you be using a cheaper knockoff watch or they're also cheaper cardiac knockoffs that you can buy on Walmart, walmart.com, they're available for $45. So that's a problem. And then there's the age issue, right? A lot of our patients are elderly and they, when they come in, they actually do have an Apple watch on, but they don't know how to use it. And you say, Hey, which version of the Apple watch is it? And you have to go down the whole thing. And you, you want to do the best you can for your patients, but time is running out. You're looking at the clock and there's 20 people waiting outside. So unfortunately, something we're, a lot of us are at breaking point and something needs to give. I think we have three more minutes. So just a few comments, last comments on legality. So if someone brings us the data, you look at it, you make a clinical decision. Any comments on if we missed it, or let's say a patient had a stroke and we missed the AFib or whatever, and how is it, does it hold in the court of law or can we be held legally, you know, legally responsible for the decision we made? Any comments? Usually it requires negligence, right? And a duty to care and negligence. I wouldn't say that misrequire, meets that standard and most JD points of view, but I mean, even legals aside, we, we don't want it to happen. Right. And so people are flooding you with the things I've had strange edge cases where somebody, I thought it was the person was atrial fibrillation and they were letting their friend use their phone who was in atrial fibrillation because they just wanted to see that it would get the reading, right. But it set off all the alarm backs on ours. When we're, when we're doing some of the monitoring. So I, I, I don't think it's, I don't think there's a high legal risk. I haven't heard of any cases of people being sued for things like that, but at the same time, again, we have to deal with other issues to make sure it doesn't happen. I, you know, I definitely think it's a risk that if you have like a patient who has a stroke and then someone digs through the HR and there are a couple Apple watch tracing showing atrial fibrillation that have been sent to your staff, you know, as messages, you know, saying I'm having palpitations, there's a risk of that becoming a legal issue. And, you know, it would be a standard for a court to decide if someone went that route. But the most important thing is that, you know, we don't want to miss those things, you know, regardless of what kind of legal, you know, situation it might set up for us. We want to be able to handle it efficiently and recognize that it's there when it's actually a problem. And so we want to put things, you know, put these tools into place that allow us to handle it efficiently and find the urgent, the urgent tracings and, you know, ignore, you know, for a time, the non-urgent tracings that just needs to happen more efficiently. You know, it probably is a legal risk, but I don't think it's a really, really high one. It's more that it's the right thing to do to not miss those things. Yeah. All right. I think we are pretty close to the time. I love the optimism you bring. And, you know, there's so much dialogue in this space and truly appreciate everyone's time here today. Thank you.

AI integration

wearable health technologies

data accuracy

patient education

clinical workflows

socioeconomic gaps

legal considerations

data overload

healthcare delivery