Good morning, everybody, and welcome to our session on ambulatory monitoring. My name is Andrew Crum. I'm an EP from Vancouver, past president of HRS, and a longstanding loyal person in the sector of remote monitoring. I have the privilege of having Sanjeev Narayan with me today as co-chair. And so I think without further ado, Sanjeev, do you want to introduce the first speaker? Yes, love to. It's a really, really pleasure. Very hot area. It's going to be a great session with some really great speakers. Our first speaker is Elaine Wan, who's going to be talking about a technical overview of ambulatory monitoring. Well, thank you so much, the organizers of HRS, for giving me the opportunity to present. And here I was tasked to talk about an overview, just to set the stage, no pun intended, for ambulatory cardiac monitoring. And just a quick introduction for ambulatory cardiac monitors. And I think that one of our leaders of HRS, Sana Al-Khatib, in a group had already sort of had an idea that the evolution of ambulatory cardiac monitoring was quickly evolving. And so this was a quick review that was published in JAC in 2020 about the different patches and also the influx of wearables, including the Apple Watch, Cardia, different types of checks and sort of what it could possibly do. And there's different platforms for this. And Sanjeev and I had been on this paper about the Cardiovascular Digital Health Journal about the different wearables that were invading into the consumer market that provided not only electrocardiogram, but also could do different things like blood pressure, pulse oximetry monitoring, as well as activity and using different methods like EKG leads or PPG. So just to talk about what are the benefits, well, when you have more monitoring other than a Holter, which might be one day or two days, you can also do things like a ZioPatch or any other equivalent of long continuous monitoring up to 14 days, or you can have a sort of a mobile telemetry like MCOTS. And then there are other things like Apple Watch or cardiac ECG checks. And the benefits of these are obviously that duration, long duration of monitoring, for example, for the cardiac monitors that can go up to 14 days other than an Apple Watch might be a short acute time depending on whether or not you're using a cardiac for a detection of 30 seconds or if you're continuously monitoring for any rhythm disturbances. However, there are some issues about when you're using just a single lead check versus three leads or if there's more leads or if you're just using this, for example, just using PPG waveforms, whereas you're actually looking at the rhythm itself. So this was an article that was published in 2024 just talking about evidence and utilization. So in comparison to different types of long-term cardiac monitors such as iRhythm, what they did to get out of this slide is really that the new diagnosis of specified arrhythmias such as atrial fibrillation was highest in really long-term continuous monitoring. So to really no surprise, the longer you monitor, the more likely that you're able to catch a patient who might be symptomatic from arrhythmias to find the cause of their arrhythmia or if they have a high risk factors or a high CHAD score that you might be able to find their underlying arrhythmia that they might have. And that definitely longer-term monitoring such as these long-term continuous monitoring can derive a better diagnosis than, for example, a Holter which is like 24 or 48 hours compared to short-term monitoring. Some of these limitations that we have is, number one, limited wear time. So even though two weeks is longer than 24 or 48 hours, in the lifetime of the patient on which they have these symptoms, which may be once every three weeks or four weeks, it's really limited recording time. So definitely longer recording time would have us a longer time for diagnosis of whether or not the patient has an arrhythmia. Probably to no stranger to us, our patients might complain of compliance issues with having the monitor on, even though over time, the size of the patches have decreased. And definitely patient convenience and comfortability is definitely on the radar of these companies when they manufacture these devices. And also the accuracy, which may be limited by noise or artifact, for example, if there are during exercise or activity. But overall, the data that's provided may be limited because you're just focusing on the arrhythmia or the ECG, whereas there's a plethora of data that we may be interested in. So other evidence and utilization of the ambulatory cardiac monitor shows how can this actually affect clinical outcomes? For example, reduce inpatient hospitalizations or emergency room visits or outpatient visits. And basically, it shows that all this type of long-term cardiac monitor seems to outperform Holter monitors or short-term monitoring. So of course, for us as cardiac electrophysiologists, the other option that we have is implantable loop recorders. Of course, it's a little bit more invasive because it's implanted underneath the skin. But there's a larger future for implantable long-term recordering. And this is an example of something, a proposed dashboard that we set up in the Cardiovascular Journal Health Journal, which is not looking at arrhythmia such as electrocardiograms, but also perhaps their exercise, their sleep trends, perhaps their weight, and all this cumulative data that we might get from ambulatory cardiac monitoring beyond just cardiac arrhythmia. So, just to magnify this a little bit more, there are upcoming general health trials. There's the LexDX Trends trial, which I have a poster today at 1 o'clock if you want to stop by in the poster hall, and where we put in a loop recorder with the Boston Scientific HeartLogic monitoring system sensors that look at, for example, impedance monitoring, activity, respiratory rate, the angle at which they sleep, and how that augmented data, sensor data, not only cardiac rhythm, may help us to diagnose things such as heart failure decompensation events. And also on their ongoing trial that we're looking at that we magnify to not only U.S. sites, but also European sites, how we can use smartwatch and machine learning to not look at just atrial fibrillation, but other things like SVTs or PVCs to help understand the rhythm burden in patients. So the last slide I wanted to talk about is what is the future of research in ambulatory cardiac monitoring? Well, I think the main question that we have as electrophysiologists still is, what is the best way to measure AF burden? And what if there are symptoms or if the patients are asymptomatic? Is really the best measurement of percent AF over the amount of time that we're monitoring? So of course we established in the previous slides that the longer we monitor, the better we are in getting a diagnosis. But is it really just percent AF burden? And probably when we interrogate devices, we have patients that have AF for five minutes or one hour. And physiologically, the longer you have AF, we think that they have higher risk for stroke. But how does that compare to percent AF burden? So if you have short bursts versus whether or not you have small burden of long bursts, how does that compare? So I think we definitely need to come up with new ways of scoring that and analyzing that. And what is the percent AF burden, which is clinically significant for each patient? And how do we measure that before ablation versus post-ablation? So after we do an ablation, what is clinically significant in the first few weeks or the first year? And overall, probably in the upcoming years, and what has been a hot topic in this conference right now, is the use of artificial intelligence to improve diagnosis of arrhythmias, improving accuracy, and reducing false positives in these patients. And how can we use this data to personalize and tailor AF ablation, AF management for our patients? Thank you. Please feel free to ask questions through the app, through the phone, or go to the podium. I have a quick question, Elaine. And it was great talk. You mentioned duration, and as you record for longer, the likelihood of picking something up is greater. On the other hand, even that's probabilistic, unless you have a continuous recording all the time. Flip it, you can do AI on a 12 EDCG. But that doesn't always hold. The Mayo data doesn't generalize in all studies. What's the sweet spot? Yeah, I think those are great questions. For example, like you mentioned, if you do an EKG on a patient in normal sinus rhythm, can you use the P wave to predict their likelihood of going into atrial fibrillation? And there's a lot of question is, what data can you derive from a single lead versus a 12-lead monitor, and for how long? And I'm going to double down on this and say, what if you don't just take the EKG, but also take other things like imaging information, like an echo and MRI to further risk stratify? So I think in this age that you're leading, Sanjeeva, this AI and digital tools, when we incorporate more of this information, not only electrocardiographic information, but also echocardiographic information, and also like these sensors, which are giving us information about activity, impedance, heart failure. Maybe perhaps all this information can give us some more optimization of what we can predict about our patients, and that is it only AF duration, but all these other things that we have to look at that can improve risk stratification for our patients. So I have a practical question. Thank you very much. So I saw a patient a couple of weeks ago in AFib clinic, 66-year-old guy who had an ablation two years before, no clinical recurrence, and twice-a-year normal halters, who wears an Apple Watch, who says, won't the Apple Watch tell me if I get AFib? And then he, because he'd had a lower GI bleed, not a bad one, but he had a bleed. So he was a good case to say, CHAT score of one, stop the anticoagulation, use the Apple Watch or a smart watch or some system to be a safety detector. Right. So I guess, obviously, from this, I'm going to answer this that it's definitely a patient-physician discussion about shared decision-making about how to proceed with this. I see this sort of a knife that cuts two ways. Definitely as physicians, we often get too many false alerts also from Apple Watches, and we don't know if it's true or not. So when you do a PPG measurement through an Apple Watch, how do we know it's really AFib versus PACs or versus AT, or an atrial flutter that would still mandate anticoagulation? So I think those are difficult to answer. And you bring up a point, Andrew, is that our patients would rather wear an Apple Watch than, for example, a monitor which has single or three-lead diagnosis of what the arrhythmia might be. But overall, I think, given the difficulty of this case about the GI bleed and the CHAT risk score, if it's low, then perhaps, overall, the risk for short-term stopping anticoagulation is not an issue. But it's definitely, if there's concern about recurrence and depending on, like you mentioned, he's symptomatic, but would there possibly be asymptomatic episodes that we don't know about when he doesn't feel it? I think all of that probably needs more true diagnostic data, which I think this paper that I showed that had all the data about utilization of long-term continuous monitoring versus Holter versus other methods shows that, really, long-term continuous monitoring to prove or demonstrate that the patient is in this arrhythmia or not is the best way to have more objective data. Yeah. I think, so, you know, my thought process with this guy was, of course, he was an engineer, so he's obsessively measuring everything. Of course. He said he could send me all the data I wanted. But the thought process in this situation is that it's an imperfect but very good screening tool to look for asymptomatic recurrence, and they're only going to get better. And so, given the fact he has a demonstrated downside to the continued anticoagulation, then I thought it was quite reasonable to stop it. And I'm also from Canada, where I don't get sued. So, you know, I thought net benefit. And I really like the AFib framework that they knew HA guidelines with sort of burden and duration. And then, also, the other thing was his left atrium was normal in size. So I thought the probability of recurrence wasn't zero, but it was low-ish. But I think those are important questions, because do you have a standardized approach to monitoring your post-ablation patients, and is it influenced by the substrate? Which is actually one of the questions from the floor. I mean, we, I mean, definitely, the Heart Rhythm Society is, and has in the past, talked about monitoring our patients after ablation. We don't necessarily have a standardized guideline of, like, three months, six months, or one year, depending on what the patient is feeling or what they're also agreeable to. I think all of these are great questions, and do we put monitors on these patients if, number one, if they're more, can be more amenable to using their Apple Watch, but then there's concerns about the data integrity of these things, and whether or not we're missing episodes. So I think, overall, this is evolving, Matt, like you talked, as you mentioned, the amount of AF duration, I think, needs to be better defined, what we're considering significant. Excellent. Okay. Thank you. Thank you. Thank you very much, Elaine. Absolutely. Okay. Our next speaker is Jonathan Chu. He has the distinct privilege of sleeping in his own bed. Yes. If I'm not mistaken. And looking very cool, too. Very cool. Looking very well-rested and very West Coast. I could put my, anyway, please enlighten us. Thank you, Dr. Cron, Dr. Narayan, and really a pleasure. You know, welcome to San Diego. I knew San Diego was a thing when Heart Rhythm was here, and just how people talked about San Diego going to the meeting. So hopefully you all are having that experience here in San Diego, so it's a pleasure to be here and talk to you in our backyard. I want to talk about ambulatory cardiac monitors and the importance of data reliability, accuracy, and duration. I'm going to focus on duration, but really would like to talk about the suite of ambulatory cardiac monitoring and convince you that it's essential for diagnosing cardiac arrhythmias, with the increase of prevalence of atrial fibrillation, obviously, and then syncope palpitations, specifically cryptogenic stroke, really drives the demand of these monitors and across the spectrum. And I'd like to focus really on three areas, reliability, accuracy, and duration. When I talk about the spectrum of detection in these type of monitors, I tend to be a duration person. Generally, there seems to be acuity as a way to stratify how one thinks about it. I tend to think about it in regards to duration. And unfortunately, our world lives in billing codes, right? So this is how I've a little bit divided that duration aspect of Holter when it comes to 24 to 48 hours, long-term continuous monitoring, ambulatory ECG in regards to 7 to 15 days in regards to the billing code, event monitoring and mobile cardiac telemetry up to 30 days, and then we're talking about implantable loop recorders in regards to 3 to 4 years in regards to battery life. And then we also have, it's not a billing code, but we have to think about wearables, including Apple Watch, CardioMobile. Those are really important aspects in regards to the spectrum of detection. And we just recently published a paper in regards to nomenclature in pulse field ablation. I think really, when we talk about ambulatory care monitors, we'd love to publish, if anybody's interested, a paper on defining the nomenclature in this type of monitoring, because it really needs to be an area. When we talk about clinical relevance, up to 30% of arrhythmias are asymptomatic. We know this in our everyday practice. And we also know that false negatives can really be critical in delaying diagnoses such as atrial fibrillation, VT, and pauses. And the optimal device choices really impacts patient outcomes and cost effectiveness. When we talk about the definition of the three things I'm going to be talking about in regards to reliability, I think we need to think about signal acquisition. We need to avoid dropout. When we talk about accuracy, we're talking about correct interpretation of rhythm abnormalities. And part of that has to do with automatic algorithms involved with these ambulatory cardiac monitors. Duration is what I'd like to focus on. We need to make sure that the monitoring period is long enough to capture intermittent or asymptomatic events. And let's focus on our first point, which is data reliability. I think the main thing that we need to make sure to avoid problematic aspects with reliability is minimizing artifacts. So what are those issues? Whether it are muscle activity, poor electro-contact, patient movement, these can create artifacts that obscure true cardiac signals. We need advanced signal processing and device design to reduce artifacts, and poor signal quality can lead to misinterpretation of missed events. In regards to ensuring continuous recording, now a device is not going to work unless one is wearing it. So if there's a device malfunction, battery issues or changing out the battery or patient non-compliance, then certainly we're gonna have poor reliability. So accurate data storage and transmission are also vital. When we look at these type of reliability in regards to the data, these are some of the best studies looking at the signal quality based on artifact-free time. Barrett et al in regards to Holter, we're probably getting in the 78% artifact-free time period. Stein-Hubble study, probably average signal quality is quite high in long-term continuous monitoring with about 92% artifact-free time. And then when we look at probably one of the standard of care studies in regards to implantable loop recorders with crystal AF in 2014 in New England Journal, we're talking about very high average signal quality with artifact-free time in the 98% range. In regards to accuracy, what are we really trying to achieve? We're trying to basically achieve ambulatory monitoring that approaches in-office 12-lead ECG as a gold standard, right? We really want this primary goal of ambulatory monitoring being able to classify these arrhythmias almost like a 12-lead ECG. And our expectations are high sensitivity, high specificity, avoiding false positive, avoiding false negatives. When we talk about the spectrum across the duration, Holter monitors are probably good for sustained arrhythmias, maybe atrial fibrillation, maybe getting a day's worth of what their average heart rate is or a PBC burden in one day. But these are wired monitors that only are one to two days. We know that long-term continuous monitoring is able to achieve this with longer duration and that event monitors and MCTs, although they do rely on algorithms, tend to go for 30 days and have the risk of over-diagnosis or over-reporting. Now, when we compare these across the spectrum of duration, how do they compare in regards to duration, symptom or auto-capturing, and patient compliance and diagnostic yield? When we look at Holter monitors from 24 to 48 hours, certainly there can be a diary that doesn't tend to have symptom triggers in relation to the patient aspect and their ability to interact with the device. Although the compliance is high, likely because of the duration being short, the diagnostic yield is likely quite low. When we go up to 15 days by the billing code with long-term continuous monitoring, the symptom triggering or auto-capture can be there, but the patient compliance tends to be very high due to some of the patch nature of the devices and diagnostic yield can be at least moderate. In regards to event monitoring MCT up to 30 days, the symptom triggering auto-capture tends to be there. The patient compliance is moderate with diagnostic yield being in the moderate category. In implantable loop recorders up to three to five years, certainly with auto-capture, the patient compliance tends to be very high in an implantable device with a diagnostic yield that is quite high in and of itself. Let me spend the last time of this talk in regards to duration of monitoring. I think as cardiac electrophysiologists, we realize that burden and duration of monitoring is really, really important. And let me give a base case for the diagnostic yield in any suspected arrhythmia that is intermittent to be proportional to duration of monitoring and the base case of cryptogenic stroke when that is the diagnosis and our suspected arrhythmia is atrial fibrillation. When it comes to monitoring, we do have studies in this base case type scenario of cryptogenic stroke, specifically published in the same New England Journal in 2014 in regards to ambulatory cardiac monitoring, comparing standard monitoring, which would be essentially a Holter versus a 30-day event monitor in specifically those patients that presented with cryptogenic stroke. In this EMBRACE trial in 2014, AF was of 2.5 minutes or longer was their definition, was found in 9.9% of patients in this 30-day monitor compared to 2.5% in the control group. And then in the same journal, when we look at longer-term monitoring with Sana et al looking at crystal AF, one of the first randomized control trials of ILRs versus standard monitoring detection of AF to assess whether longer monitoring can detect atrial fibrillation and duration being a thing in cryptogenic stroke, that when we look out to only six months in the initial primary endpoint of crystal AF, that detection of atrial fibrillation was 8.9% compared to 1.4% in those individuals that were in the control group. This is a case example of a real-life patient in my clinic who is a 72-year-old man who came in with posterior left MCA stroke on MRI, was admitted to the NeuroStroke Service. They did their standard protocol of TEE, hyper-COAG panel, 14-day event monitor, placed on Plavix, and then got an ILR implanted. This was the monitoring remote that we all like to read because it says everything is normal until this happens, which is atrial fibrillation six months after implant of the device via implantable loop recorder. And what's interesting about these type of devices where we can monitor for longer and remotely re-monitor, you can start at a PIXIVAN, which is what he was. He lives in an RV and actually had his bedside monitor plugged into his RV and then moved to Tucson, Arizona from San Diego and wants to be monitored by our office until his monitor can be transferred to an EP in Arizona. Just going to show that the ability for us to monitor in a long-duration-type frame is present in aspects, including cryptogenic stroke. So when we talk about duration to detection and the correlation, if we looked at cryptogenic stroke as a baseline based on some of these studies, including Gladstone, SANA, and Rafael, 24 to 48-hour Holter monitoring will probably detect less than 5% of patients with atrial fibrillation in this aspect. When you go out to 30 days, it's probably up to maybe 16% about. Based on these studies and ILRs, whether you look at crystal AF or other studies are gonna be about 9 to 20% at six months and about 30% when you take crystal AF out to three years. So in summary, ECG monitoring in regards to ambulatory care monitoring, data reliability is important. Accuracy is also important. We need to prioritize devices with high reliability and high sensitivity and specificity. And I'd like to state that duration of monitoring is really important to yield, particularly for our diagnoses such as cryptogenic stroke. So thank you very much. And hopefully you're having a great time in San Diego. Questions or comments from the audience? I might start with one and say that I recall very distinctly during the sort of break of COVID era that there was a bunch of validation studies where all of the different platforms of technologies had simultaneous recordings with validation. And then that helped, for instance, determine which had the highest performance and so on. But it feels to me like there's some reliance on AI to fix this and it's not really happening anymore except for regulatory purposes. So do we have preferred platforms even though we don't wanna talk to commercially? Like accuracy wise? Did you pick a vendor because it's the best? Do you pick a vendor for your patch after AFib ablation because it's the cheapest or the best tolerated? Dr. Crown, I'll speak to our institution and what we do. We tend to be not one vendor across the platform, whether it's CRM, ablation, mapping, and ambulatory cardiac monitoring. I think when it comes to algorithms or I guess you're speaking to reliability and accuracy of a specific vendor versus another, I'm not certain that that's the only variable involved. There's so many aspects across this in regards to what is the patient compliance, whether or not they're going to wear the device and whether it's in patch form or needs to be an implantable loop recorder. So in general, I would say we use all vendors including those that have patches in event monitoring MCT. For us, we tend to choose based on duration and what is needed in regards to that. And then we tend to have all types of monitors including all vendors for implantable loop recorders and MCTs as well. While people are coming up with questions, I had one, it was a great talk, John, really thoughtful. You look at the detection rates of monitoring after a cryptogenic stroke and it goes up, but still it takes a heck of a long time to detect, even in crystal, it took months. You could easily say, you know what, the AFib's an epiphenomenon, the old argument, it's really, so what else should we be looking at? How do you take that into account in your patients? Do you look at, do you even consider doing an MRI to look at some area of atriomyopathy, CHADS, VASC? And then quickly, you didn't touch on watches. A device, our patients are using that a lot. So any thoughts on those issues? Yeah, to the first point, I would say that was the question about ILRs, correct? It was the notion, is that one episode of AF three months or six months after the cryptogenic stroke, is that what happened before the stroke? Is the AF the key, or is it the atriomyopathy? And that, of course, is key for REACT. Yeah, as you know, Dr. Narayan, with stroke AF being about the same rates of detection in a small or large vessel, the stroke population, you wonder if it is an epiphenomenon, as you state, because that detection of atrial fibrillation up to three years is about 30% as well compared to crystal AF. So is it really cryptogenic stroke? Is it any stroke? Do we detect AFib in a CHADS, VASC high-score population? And I think that that data, combined with crystal AF, would suggest it. So I agree and assent to what you're saying, which is, I think this just tends to be, AFib is prevalent, and it's a high-risk population, and we're gonna detect it if we look for it. So hopefully that answers that question. In regards to the watch and wearable phenomena, I think it comes down to a billing code, right? I'm sorry to go back to billing, but I don't tend to prescribe watches or CardioMobile. My patients tend to prescribe it to themselves. I think if we are able to digest this data, make it billable, and also able to be in our EMR so that we are responsible for this data in that type of environment, I think it becomes an aspect, as I showed in regards to the spectrum, that we really consider in our armamentarium of ambulatory cardiac monitoring. And I might just add, as somebody who practices in a second-world country, that I... Hey, but you're part of us. Yeah. That's right, the 51st state. It's not a state, yeah. So I go up north five weeks a year to some people who choose not to have a phone and live in the bush or live in an area where it can take three months to get a holter. And I advocate for watches and wearables because it's better than nothing, even though it doesn't create a billable encounter or something like that, because otherwise it's pure judgment or delays, or they have to travel 200 miles to go get the holter and then do it again the next day. You know, it's crazy. But that's the reality of some, either both health systems and also where people live and their access issues. So if you asked me, I would say that it's quite realistic that this is a rapidly dynamically changing thing. And that in the end, reimbursement should influence but not guide. I totally agree with you, Dr. Krohn. I think reimbursement, although it shouldn't drive the show, does unfortunately do something. And then I do think another aspect perhaps, which is integrating with an electronic medical record and making that not only documented at least, would be an important aspect specifically with the wearables. There's actually one quick, maybe just one quick audience question that's online. Sorry, Elaine, but just this one. Is Crystal AF, the standard of care, sorry, in was a 12 lead ECG clinic visit. Would it help to put ILR against the continuous non-interrupted monitor to determine if the captured events were similar? So if you could answer that quickly, John. That's a great question. Our neurostroke colleagues who I interact with a lot ask the same question, right? Embrace was in the New England Journal in 2014, so was Crystal. Do we need a head-to-head with 30-day versus ILR? I'll tell you what they're doing. They're struggling and we have some data that suggests when the patient is actually able to choose that our neurostroke colleagues at our institution actually have them choose based on the Embrace and Crystal AF, what they want, that they feel like their stroke care is better as a result of choosing between these two trials. I don't know of any head-to-head trial either in clinicaltrials.gov or ongoing looking at that specifically, but I do think it is an area of interest. They tend to get a 30-day monitor and then if that's negative, put an implantable loop recorder, but that's never been studied in a true randomized control trial fashion. Dr. Wong. Yeah, no, just because we're talking about Crystal AF or the trial that I was alluding to that I'm presenting the poster on today for Lux DX Trends which had over 250 patients, the risk factors that we usually think about like hypertension, congestive heart failure, diabetes, I feel that's changing all the time because for our trial, we're looking at preserved EF or moderately reduced EF and we found in 12 months that 25% of people had newly diagnosed atrial fibrillation. So what I'm trying to say is that in Crystal AF, we're looking at people, we're finding cryptogenic sources of stroke, but the risk factors are constantly additive or changing over time as we get to do more research and using these loop recorders to better re-stratify our patients and find out, expand what may be the risk factors that may lead to stroke due to atrial fibrillation. Then I, did you have? Yeah, I was just gonna say, HFPEF is sort of also uncharted waters in that space too, right? So that segues quite on to another couple of questions from the audience. What would be an interesting parameter besides ECG to monitor? I'm sure there's opinions on that. I think what we're getting at and what Dr. Wan is also saying is not only perhaps comorbidities are improving, but other aspects of vital signs. I know certainly some of the companies involved in ambulatory cardiac monitoring are looking at different vital sign surrogates to look at besides ECG. I guess I would say a couple things that we found to be helpful either in ILRs or whether it's the patient with POTS or whatever, so position or activity I think is helpful. So I would say activity seems to be something that we're utilizing whether across the spectrum with ILRs or other ambulatory cardiac monitors. I don't know if you all have, obviously pulse ox was important during COVID, but maybe not as much nowadays, but I think activity or position seems to be what we use the most. Given the average BMI, maybe it's sleep apnea. Sleep apnea is good, I think. I don't know if Mintu's in the audience still, but I know that there was a talk I heard him give where there's interest in combining a lot of vitals. If you want to. Yeah, you can step up to the mic, but I don't think it's working, so you can just shout. We'll have Mintu talk about it. I was kind of thinking, I was alluding to maybe Mintu saying something in regards to vital sign monitoring. I'm conflicted as an employee and corporate officer of iRhythm, but I think one interest, and I don't want to cut into Sunit's time, is the ability to think of extended indications outside of core ECG. Core ECG in our world as EP is super important, but when you get to the ability to pre-diagnose sleep, apply AI for ejection fraction, embed accelerometer data for activity, think of a broader vitals platform that may have use cases outside of sort of core arrhythmia surveillance, you can open up a lot of opportunity. Thank you. We have 15-minute panel discussion built into the program, so that's why we're not cutting into Sunit's time. We're inviting him to the podium. Thank you very much, Jonathan. Yeah, we had to clarify that, Mint. Thank you very much. So it's really a pleasure to introduce our last speaker of the session, Dr. Sunit Mittal, who'll be putting all of this into a practical real-world context. I don't know if any of that is true, Sanjeev, but thank you very much. Okay, I'd like you to start by imagining the walk many of you will take of an hour from now when this session ends, downstairs to the exhibit hall, where you are going to be inundated by a whole number of ambulatory external ECG companies. And it is very easy to leave that walk thinking that this is a completely commoditized space and that maybe all that really matters is the duration of monitoring. And I think in the next 15 minutes, what I'd like to argue is maybe we need to think about this a little bit differently. And maybe what we need to focus in on, does our choices impact clinical outcomes and importantly, healthcare costs? And maybe that is the variable that should help inform choice. And maybe that involves something more than monitoring alone. Now to set the stage for this, I have to level set a couple of things. The very first thing is, if I can operate this unit, is what are we talking about today? So for the purposes of our talk, this was titled ACMs, Ambulatory Cardiac Monitors. And I'm going to say that for definition, we're going to limit it to electrocardiographic monitors. And so here, you've seen the prior two speakers really highlight our evolution from the standard 12-lead CCGs to leaded Holter monitors to now patch-based technologies for long-term continuous monitoring or even mobile cardiovascular telemetry monitoring. Andrew has, of course, raised watches and phones, and there's implantable devices. In the interest of time, I'm going to level set by saying I will not be addressing watches and phones today, but I will be addressing the other forms of ambulatory electrocardiographic monitoring. Now Jonathan points out very well that diagnostic codes are important, and this will be to some extent a United States-focused talk, because we're very fortunate in that we have a number of distinct diagnostic codes that allow us to think about these technologies differently. So we have codes for standard Holter monitoring, but recently we were able to get codes for more extended duration of monitoring, and that includes Holter monitoring out to seven days, Holter monitoring from one to two weeks. We even have the ability to get mobile telemetry or autotriggered modes that go out to 30 days, and of course there's implantable loop recorder monitoring, which can go out ad infinitum. And this then allows us at least to inform some choice, but often what we really think about is, hey, if I need 30 days of monitoring, which is the most I can go, that informs one choice of technology, which is typically some form of mobile cardiac telemetry monitoring, and if it's a lesser amount, we think of long-term continuous monitoring with a Holter as being limited to just about two weeks of time. Now often that is enough, and we certainly, I think, know as electrophysiologists and increasingly we're able to convince our cardiology community to move away from the 24- to 48-hour model, because we've recognized the work that Mintu has done when he was on the other side of life, showing that really one to two days of monitoring is just not enough to get to diagnostic yield, and that really as no surprise as you start to monitor longer, the diagnostic yield goes up. And I think certainly electrophysiologists have bought into this concept and increasing cardiologists so that seven to 14 days of monitoring has really become the standard. Now besides that, I think we have to really think about the other variables that go into this. As you think about this spectrum of monitoring, the things that I think we're integrating every time is, hey, are we monitoring for symptomatic arrhythmias or asymptomatic arrhythmias? Of course, for symptomatic arrhythmias, you may need to monitor a little bit less than you would for asymptomatic arrhythmias. Is it a known diagnosis, or are you trying to exclude something? If it's a known diagnosis, I may need a little bit less monitoring to confirm some aspect of the disease I'm looking for, whereas if I'm trying to exclude atrial fibrillation in someone who may have had a cryptogenic stroke, for example, I'm going to want a longer period of monitoring. And of course, what is the urgency for notification? Am I looking to be awakened at 1 a.m. with a call about something that was found on this, or am I willing to just aggregate the data and think about this at one time? Now it turns out that the implications on healthcare costs and utilizations for these decisions that we make every day really wasn't very well delineated until we had the ability to work and publish this study called the Camelot Study. And both the prior authors or speakers have spoken to this study, but I want to just take you through this so that you understand a little bit of what it is that we were trying to do and how I believe that this kind of sets the field for how we should go in the future. So this was, first of all, a very large study. We started with nearly 1.6 million people. It was a retrospective study looking at a Medicare fee-for-service sample accumulated over nearly four years of time, and these were basically patients who were being subject to ambulatory cardiac monitoring with four strategies that I'll outline in a second. And we basically followed these patients to try to understand some impact on more heart outcomes other than just what we think about normally, which is we made some arrhythmic yield, which is, of course, important but may not be enough. So the types of monitors we included were Holter monitors, what you think of as one-to-two-day Holter monitors, longer-term Holter monitoring, three to 14 days, external patient-activated event recorders, which at the time of the study were still used or less used now, and mobile cardiac telemetry monitors. And again, it's important to know, at least in the U.S. where this technology exists, it's considered to be synonymous with this ability to monitor out to 30 days. We wanted to see whether there were differences in arrhythmic diagnosis by looking at the things you would be interested in, AFib, flutter, SVT, VT, AV blocks, sick sinus syndrome. And we looked at three clinical endpoints. One is, within 90 days of that monitoring, what was the likelihood of a diagnostic yield? Within 180 days of application of that monitor, what was the likelihood a patient needed another monitor? And of course, what were the annualized healthcare resource utilization at that time? Now, again, a very large study, but again, speaks to the fact that although we have moved on as an EP community to things outside of Holter, at the time the study is done, again, look at the distribution of what's being done. Almost half the patients are getting just standard one-and-two-day Holter monitors. Of the remainder, still a large use of MCT monitors, which I think reflects the fact that we have bought into the fact that duration of monitoring is so critical, and that gives us the longest duration of monitoring possible, and lesser use of Holters and patient-activated recorders, and the demographics you see here. Now, what's really interesting, and I think novel about this study, is the ability for the first time to really differentiate the impact of the choice we made at the time these monitors were acquired, and on these outcomes that we studied. And the first one that I'm showing you here is the diagnostic yield for a specific arrhythmia. And what you see is that relative to a reference case standard, in this case a Zio XT Holter monitor, you see that the other forms of monitoring, even when it was greater durations of monitoring, like with a mobile telemetry monitor, had lower amounts of diagnostic yield. And even within the Holter monitor areas, there were differences in diagnostic yield, suggesting that this may not be as commoditized as we think it is. Another end point, of course, is you certainly don't want to put a monitor on someone and have them need another monitor because you needed more diagnostic information. Again, you see the same patterns of finding in that relative to other monitoring type, this extended Holter monitor had the least likely need for retesting. And finally, Elaine alluded to this, annualized healthcare resource utilization, we're looking at three variables, need for inpatient hospitalizations, ED visits and outpatient visits, again showing the same signal. We have to now start to think about why is it that this is happening and start to answer some of the things that Andrew was asking about before. Because as clinicians, we just don't get insight into this information because while companies submit this information to the FDA, presumably, we don't see this in a comparative form. So are the differences because there's differences in wear time between these things. Maybe patients perceive one device is more comfortable than another. Maybe the addition of one is better. Maybe compliance related to the need for path changes or battery charging is we can hypothesize but we don't know. Maybe the signal quality is simply better between one system or another. Maybe the algorithms, the so-called AI algorithms that are being run in the background, one is better than the other. And maybe the data reporting, including the technicians, is better or not. All things that we have to get to the bottom of. Now at the same time we were publishing Camelot was this study that I want to highlight from the EP group at Penn, which I think is a remarkably powerful study. And in this innovative study, what they did was they took patients with atrial fibrillation and they simultaneously placed long-term Holter monitors and mobile cardiac telemetry monitors at the same time on the same patient. And what you see here are a number of patients, maybe difficult to see, where you see that there's complete disparate information with the MCT monitors grossly underperforming a Holter monitor. So cases where a patient was in persistent atrial fibrillation and MCT monitors suggesting that patients were in paroxysmal atrial fibrillation and completely underestimating the amount of atrial fibrillation. Why is this important? Because it's important because the choice of technology may be more important just than the duration of monitoring. Now monitoring is important and that gets into this whole issue of implantable loop recorders and when we should be using them. Now I can't do a whole sermon on loop recorders, but I can focus on one use case scenario. And I'm going to, today for the purposes of this talk, focus in on patients that have known atrial fibrillation and let's hypothesize these are patients that we treat every day that we've ablated for atrial fibrillation. And I want to talk about why we may want to extend our monitoring to a longer duration. Well certainly on a macro level, we want to define their symptoms, is it related to atrial fibrillation or not. You may be using the information to maximize medical therapy, antirethmic drugs, anticoagulation. And whether it's your own knowledge or it's part of a clinical trial, ultimately it is still the gold standard for determining whether what you did was successful or not. Now my favorite study, one of my favorite studies of all time is this study and this is why I think that we're challenged every day and why we make the wrong choices. So this is the concept of AF density and AF density is an important concept because it really tells us why we get challenged with the yield of these monitors that we place externally and why it's important is because AF density is the one parameter you don't know how a patient will react. So those of you who have not seen this figure just want to take you through this for one second. So on the left side of the panel are two patients, patients in red and a patient in blue. Both have an implantable device that measures their AFib and over the course of a year, both have exactly the same AF burden of 28%. The difference is the patient in blue has high density atrial fibrillation, which is they had this one prolonged episode of AFib, the rest of the year was sent in sinus rhythm. The patient in red had ongoing episodes of atrial fibrillation, you know, very frequently, but none of them was a persistent episode in the sense that it was like any given day they had some atrial fibrillation. Well the reason it's important is that if a patient has low density atrial fibrillation, you're going to do great with one short term or two short term external monitors. But if the patient has high density atrial fibrillation, you're likely to miss this completely. So not knowing the density of atrial fibrillation, it really mucks our ability to really understand what's going on. The other thing is we're increasingly realizing that you need patient information and AFib is well beyond a year, which is really where the guidelines speak to a year and then they're silent to it. In our lab, we've been very interested on understanding the implications on outcomes when you follow patients out to a year. When Dr. Jonathan Steinberg was working with our lab, he coined this term called very late recurrence of atrial fibrillation. And what very late recurrence of atrial fibrillation is you ablate someone for AFib, you follow them for a year like most guidelines do, they're completely free of atrial fibrillation. Most people basically then discharge them for their clinic. And we said, we're now going to start following these patients to see what happens to them over time. And you see, not surprisingly, patients develop atrial fibrillation or very late recurrences. And you will not have this information if you don't have longer episodes of monitoring. More recently, my colleague Dan Musad, working with Nicole Milstein with continuous monitoring, showed that this phenomenon exists even in a more modern age. And again, you follow these patients, especially those with high CHADS VASc scores, it is not surprising that many of them will develop very late recurrences of atrial fibrillation. Here's another way why this is important. And this is our attempt to see how we can get misinformed when monitoring is of short duration. We published data from our lab that have been presented again by Dan Musad and colleagues where we looked at the impact on body weight on what happens after afib ablation. And here we're stratifying AF outcomes with continuous monitoring with a loop recorder, depending on whether your BMI is greater than or less 30. You see, if you've truncated it a year, you would walk away with the impression that body mass index makes no difference on outcomes. But if you start following these people longer, you see, well, this makes a big difference. The heavier you are, the worse the outcomes. And having that information, then understanding what's happening to the outcomes of this patient should go back and help inform your decisions going forward. Now again, at least with ILRs, the performance data is better published than it is for ambulatory external monitors. And we know that it is a function of the duration of AFib you're looking for and also the population in which you're measuring. You see, in certain patients, it does very well. In certain patients, like cryptogenic stroke, not as well. But what we do know is that most ILRs do very well for detecting AFib episodes that are one hour or greater. And we also know that through the use of artificial intelligence, we were able to show that you can further increase the performance of these devices to be more robust. And hopefully, we'll get to the point one day where we can really take to the bank that when it says there's AFib, that there's truly atrial fibrillation. So I want to leave you with these conclusions that certainly when we started with Camelot, large contemporary analysis of Medicare beneficiaries, there was a substantial variation in monitoring strategy. I would argue when there's a lot of variation, that's synonymous with we don't know what we're doing. Every time we know what we're doing, there's a more homogenized approach. We're just not there yet. Use of long-term Holter monitoring had the highest diagnostic yield, which may be very surprising to you. It was associated with the best diagnostic yield, the lowest need for repeat testing, the lowest need for subsequent hospitalizations. Certainly the role of ILRs continues to evolve, certainly with the rapid development of wearables. But I want to leave you with that different monitoring strategies produce different results with respect to diagnosis and outcomes and care. And this is what we should be focusing on, not just the duration of monitoring alone. Thank you very much for your time and attention. Great talk, Sunil. Really lovely. Thank you. There are a few questions online. If anybody has any others, please come to the microphones. First one, so first of all, I was going to ask you specifically, let's do this. Do you think your clinic support can yield better outcomes? So the clinic support per se. For instance, are telemetry reports being reviewed at the end of the service or as alerts are notified? How do you manage it? So I'm assuming, Sanjeev, that question relates to MCT monitoring, which produces daily reports and end reports. I think that in our practice, these things are reviewed daily as best we can. I want to be very transparent to people. After seeing the results of Camelot, we have really shifted our practice away from MCT monitoring. So we use very little of it. But despite what I think all of us on the podium, it's easy for me to say that, oh, we look at them twice a day, three times a day, each and every day. But the reality is the objective, unbiased data show that most people do not look at this until the end of service report is available, just because we don't have the time or bandwidth to do so. You could accelerate that with AI tools, but that's another discussion. Previous ACC consensus statements mentioned that the longer period of time before finding AF in cryptogenic stroke patients, the less likely AF was likely to be the cause. In other words, it was an epiphenomenon. What are your thoughts on that? You know, Sanjeev, as you know, all we know is that when you implant loop recorders in these patients, certainly 30% of them will have some AFib by three years for diagnosis. I think the temporal etiology and what it means, it's hard to know, because we have one fundamental problem before that is we've never been able to show that clinical decision making informed by those findings actually changes the natural history of the patient. And I think that's the more important question, rather than, without understanding that, I don't know if I can answer whether it's an epiphenomenon or not. Yeah. I mean, there's another question. I'm going to leave for a second. I'm just going to pick up on that. So, without getting into pro or con, the thoughts on REACT and the notion of really using the data from a monitor, real time, any monitor, to truly inform therapy. Yeah, thank you. So, REACT, for those of you who may not be familiar, is Dr. Passman's randomized study, smartwatch guided, pill in the pocket oral anticoagulation. It's a randomized to Apple iWatch versus standard oral anticoagulation. And in this study, you would otherwise be taking oral anticoagulation if your watch shows no atrial fibrillation. You come off of oral anticoagulation. If you're notified, based on PPG, not ECG, that you are in AFib, you go back on oral anticoagulation. You continue it for a month. I think that Andrew raised some really important issues with watches and why it's important. But I want to caution people on one very important thing, and that is that the watches being used in REACT are not the watch you pick up at the Apple store. The watch that you pick up at the Apple store has been tuned for specificity, not sensitivity. They don't want people to be told that they have atrial fibrillation unless they're absolutely sure they have atrial fibrillation. In the study, the watches and the algorithm has been tuned for sensitivity at the expense of some specificity. And so we have to keep that in mind. So if the study is positive, then we go in a direction, but with a different watch than you pick up at the store today. It's very important. Okay. Well, on that happy note, I'm going to close with torturing you with a story. So I started my fellowship in 1994. By the way, you tortured me with the story when you were sending people to the bushes for five hours or something. So 31 years ago, I started my fellowship. And on the first day, George Klein said to me, we have these new syncope monitors. We've just invented them with a Medtronic engineer. Would you like to put the first one in? And I said, George, I'll wash your car. Of course, I'll put the first one in. And so I put in the inaugural loop recorder, which was a dual chamber pacemaker generator with two electrodes on it, and it was magnet activated. So the historical perspective is just very nostalgic for me to see the evolution of this and the degree to which it's really patient focused when it comes to the spectrum of options and matching, the precision of matching the patient's clinical condition with the duration of monitoring is really so much patient facing in terms of benefit. So I think it's been a great session. I've learned a lot, and I hope the audience has too. And I thank our speakers and wish everyone a good Congress. Thank you. We're at time.

ambulatory cardiac monitoring

wearables

ECG

atrial fibrillation

data reliability

Holters

Camelot Study

AI integration

digital health