Hello, everybody. Thank you for joining us today for this session. It's digital health in EP and beyond. We have some great speakers today, so very excited about it. If you haven't already done so, you can download the HRS mobile app or scan the QR code that pops up here between talks. That will prompt you to be into our Q&A session, and you can upload your questions as we go on. I'm Lisa. I'm from Wash U. I'm a physician assistant in pediatric electrophysiology, and my co-chair here is Jag Singh, and we're going to start here with DJ Lakharedi from Kansas City Heart Rhythm Institute, and he'll be talking about crossing the chasm from ED to EP using digital health tools to decrease hospitalization. Thank you, Chairman and Chairwoman and colleagues here. These are my disclosures. So when you really look at healthcare models and care delivery-related issues, there are several ways you can really tackle the gaps that exist in a patient flow. So if you really take atrial fibrillation as a model, I think it's actually a very perfect model to study the nuances of care process flow and how technology could be leveraged and really creating and improving access to care-related issues and improving outcomes. So let's get on with that. So atrial fibrillation, as we know, is everybody's bane of existence. It impacts anywhere from six to seven million now, and in about another four or five years, this number is going to hit over 12 million people with a tremendous impact on clinical morbidity involving strokes, heart failure, early onset dementia, sudden chronic death, compromise in quality of life, mortality, and then, above all, the rising healthcare cost of these things. So for many years, I think we as physicians were always trained in thinking about the sort of the molecular tissue or organ-level pathophysiology, and a vast amount of the effort that goes into this is primarily on finding solutions at a molecular level or a tissue level or at an organ level, depending on whatever it is that you want to use, right? That includes anti-arrhythmic drugs or the nanotherapy or the gene therapies, understanding the characteristics of the tissue, the ablation procedures we talked about, and all of that. And when you really take that up a little, a few thousand steps up, then you have the whole autonomic nervous system modulation technologies and themes that really come into play, and atrial fibrillation perhaps is actually a very good model for autonomic modulation and its impact and its role on therapeutic aspects of how we can take care of diseases. And then when you really take it up to a little higher level where cerebral modulation is something that to think about. Then in all of this, nobody ever teaches us how do we really work on process flow changes, or we don't really focus as much on risk factor modification, or how do we really break down the hurdles that exist in access to specialized care, and how do we really leverage the existing diagnostic point-in-time diagnostic tools or continued dynamic diagnostic tools for chronic disease management. So when you take a patient's journey with atrial fibrillation, it may start in the ER or it may start in the outpatient clinic, and then from there, for any of these patients to really get into the hands of an electrophysiologist, it's a very arduous path, right? More than 70% of these patients with atrial fibrillation end up in the emergency rooms for medical care, some with new onset, and some with non-AF with related complications or other associated issues. There is a tremendous amount of heterogeneity and associated complexity in the course taken by these patients by the time they end up in the definitive treatment options. So oftentimes, these inherent delays in the referral process of atrial fibrillation can result in progression of the disease, worsening, remodeling of the pathophysiologic aspects of the disease, and then the appropriate health care providers sometimes are in shortage in certain places, right? So we thought and figured out that the emergency room is a missed opportunity to optimize atrial fibrillation care. So how many times have we not seen a patient that goes into the emergency room and gets seen by an ER doctor. They either get discharged from the ER after they get started on oral anticoagulation, their antirrhythmic drug or an aminodal block are adjusted, and they get sent home with really no defined or foolproof follow-up with their cardiologist or an electrophysiologist. So they were asked to go back to a primary care physician. The primary care physicians in this day and age, right, when the entire internal medicine workforce, most of it goes into subspecialties and the other half goes into the hospitalist medicine, and the true percentage of patients who actually practice outpatient-based internal medicine or family practice is abysmally low. I mean, I can't find a primary care for myself over the last 12 months, right? I mean, I have to pull my weight and say I got to go talk to one of these hospital, the primary care physicians to actually get myself into a primary care physician's office. And imagine the plight of a poor patient who has no connections, no weight in any one of this to basically go find themselves a primary care physician. That takes about nine months. By the time the patient gets into this primary care physician's office, they've already been back in the emergency room once or twice, maybe a TIA, maybe an episode of heart failure, and they're not complaining with their medications because the ER doc only gave them the script for an oral anticoagulant for 30 days. And what does the patient know? The patient took the anticoagulation as prescribed by the ER doc for 30 days. The AFib comes back, and then they had a stroke. This is the routine story of a patient with atrial fibrillation that goes through a health care system. Yet, we are the highest spenders of health care dollars, and we have one of the worst processes that are in place because nobody is thinking about these processes. We have very, I would say, broken down electronic health record systems, and there is no continuity of care at any one of these things, right? So these variations in type of therapy that we offer and the timing and the need for initiating oral anticoagulation, ER versus inpatient therapies are quite different, and the quality of care patients get is quite different. And then, above all, nobody cares to educate the patient in this whole process, right? So all of this leads to a tremendous disparity in patient care. So we hypothesize that initiating an appropriate, standardized referral process involving an electrophysiologic service for further management of atrial fibrillation could significantly improve access to care and thereby fibrillated complications and hospitalizations compared to the routine care that currently exists in any of these health systems, right? I'm part of a large health care system, and so it was very easy for us to basically take a couple of hospitals where we could deploy this, and then the other two hospitals where we didn't do anything with it, and we let them be, and then see how the routine care can be compared to this new pathway. When you really look at this, right, it's a very simplistic intervention, a very simple intervention in an electronic medical record that prompts the ER doctor and gives the ER doctor an opportunity to send a direct referral to an electrophysiologist, right? A direct referral to electrophysiologist, if the patient goes to the hospital, that pops up on our dashboard on the outpatient side, and essentially we get them seen within one to two weeks. If the patient is admitted to the inpatient, an automatic electrophysiology consultation along with the cardiology consultation happens, and thereby there is an electrophysiologist who actually sees all of these patients by all means. So it's a simple intervention. It doesn't take rocket science. It doesn't take millions of dollars of investment, like the way you have to invent a PFA catheter, yet another PFA catheter, or it doesn't take anything or moving a mountain or anything like that, right? So we followed all these patients over a period of 12 months, and we kept counting their healthcare encounters and all the other things that happened to these people. So basic labs were drawn. Obviously, everybody gets cardiac enzymes. They get a CBC, liver functions, and coag panel, and if they have, unfortunately, chest pain, they also get CT scans and everything else. Routine workup, rule out reversible causes, and they were appropriately evaluated for stroke risk, started on a venodal blockers, anticoagulation, antirhythmic drugs, with an exclusive visit to an electrophysiologist. So this is where Eric Przyszkowski calls this as cutting out the middleman, right? So we basically made sure you cut out the middleman, you eliminated the hurdles for a patient to get into the hands of an electrophysiologist, and we tracked these patients. The primary endpoints were looking at time to definitive therapy, like antirhythmic drugs, anticoagulation, even afib ablation. Secondary endpoints included mortality, heart failure, admissions, stroke, and TIAs, major bleeding complications that meet BARC5 criteria, a number of hospital clinic results on cardioversions over a period of 12 months were looked at. So it's a relatively homogeneous group. I mean, this is pretty standard age groups, and a lot of them were males. In fact, this study is a little bit different because we actually had more women in this study compared to other studies. Again, they are routine BMI of 30. I mean, this is what you see in the mid-midwestern. Patients are high risk for hypertension, hyperlipidemia, diabetes. So the incidence of the comorbiditive profile is pretty classic for what you see in a typical atrial fibrillation patients, and the CHARSVAS score was elevated three and above. So the time to evaluation in the ERTP group was basically one day, right? I mean, if they're inpatient, it only took one day. So there are quite a few inpatients. So the range was between one day to 24 days, and the control group, right, look at it, 328 days. It takes as much as four months for an atrial fibrillation patients to see an electrophysiologist in an integrated health system. I mean, leave alone private practices, right, in an integrated health system where we have primary care physicians, cardiologists, electrophysiologists under the same umbrella. The range was 30 to 340 days. One of these patients took one year to get into the hands of an electrophysiologist. Time to ablation, significantly lower in the ERTP. Time to initiation of the antirethmic drugs, three days versus 25 days. Time to anticoagulation, two days to 17 days, and length of stay was significantly lower in the ERTP group because once EP gets involved in it, a lot of the drama that gets involved goes away, right? These decisions that don't get made and people are just looking at each other, the decisions get made, patient has a definitive plan, they get cardioverted, they get their antirethmic drug, and they're out of the hospital in no time. And look at the outcomes, right? Number of clinic visits that happen in the ERTP group, ER visits for heart related, any heart related issues, 5% versus 10%. Number of cardioversions or heart failure admissions, right? Incidents of strokes and bleeding complications, all of these are significantly lower. So what this goes to tell you is getting a patient in the hands of a definitive subspecialist, i.e. electrophysiologist, can really change the trajectory of the patient's flow in a health system and the access to appropriate healthcare resources as a result of which the therapeutic interventions that happen in these patients actually becomes significantly faster, right? So this just goes to summarize everything that we noted in this study. So it's a win-win, not only from real hard clinical outcomes, but also a number of other things that come along with it. So obviously this is not a randomized clinical trial, and I don't know if we really need a randomized controlled trial to really establish something as simple as that. The risk of positive bias, right? I mean, this is a very high volume, busy EP practice, and we are accessible. I mean, we are readily available everywhere, but this is in the middle of Kansas City where there is no major shortage of electrophysiologists there, right? And this difference could be even more significant in less organized facilities because the difference is striking. There are more persistent AFib patients in the ERTP group. Maybe that accounts for the aggressive therapeutic approach we took in these patients, and an exclusion of patients with non-AF, because we only counted patients who were de novo admissions. So I would say if we can actually bring a few hospitals or health systems together, we should probably do this study on a much larger scale to really appreciate the global impact of what it can do. So I say we can, it's easy to conclude that this study provides evidence that having an organized pathway for AFib patients can lead to improved outcomes, and we should leverage our electronic health records and appropriate process flow creation and tweaking can actually be amazing in this. So there are a lot of groups that are working on this as a concept. You have a group out of Austin called RFMX. They have this centralized dashboard that really crawls varied electronic medical records and creates this automated process flows that help a patient navigate their care continuum in a rapid fashion so that these patients don't fall through the cracks. So thank you very much for the opportunity. All right, the talk is open for questions. A provocative talk. So DJ, that was terrific. We do the same thing at Mass General. We do directly, we have this AF pathway where patients from the ER directly go to the EP lab for a cardioversion or are taken over by an EP. But let me just throw a question out to you. It's ER to EP, but then from EP to where? That becomes the issue because a lot of these patients have comorbidities, you know, hypertension, sleep apnea, we've heard about sleep apnea issues. EP to general cardiology and internal medicine. Okay, so finding the general cardiologist now who is going to assert, if I may say so, some of that role of an internist. Why are we waiting for an internist or find an internist? Absolutely. I mean, the current generation of internal medicine doctors do not have the resources at their disposal to manage these patients. Most of the internists are on an RVU target. They're more interested in seeing a count of 40 or 50 patients. They only want to address one or two problems in a visit, right? The idea of a comprehensive primary care where they're acting as anchors for comorbidity risk reduction has gone away. That level of engagement doesn't exist anymore. So as a result of which, a lot of the baseline cardiovascular risk mitigation is now in the hands of general cardiologists. So in our pathway, we made sure that each and every one of these patients actually got assigned to a general cardiologist as well. We didn't care as much whether they had an internist or not. We hope that they have an internist, but we made sure that there is a general cardiologist involved in their care. Lisa, you have a question? Well, I was just going to continue to comment on that. I think that that opens up an interesting question of the burden after care, right? I had the opportunity to work for a virtual care center before COVID pandemic happened. And the entire purpose was, it was to follow up on patients after hospitalizations with chronic medical conditions, cardiac and non-cardiac, to decrease their overall hospital readmission and hospitalization rate. This seems like another opportunity where we could use a very similar model. It's interesting to me that that hasn't really blossomed because the need is still very much there. Right. Having a post-AF conversion, urgent care kind of telehealth strategy. Dr. Peters? Yeah. So Dr. Lacheredi, congratulations. Can I just ask you, is this done under informed consent? And is your vision for multi-center done under informed consent? Yeah. And I suspected you'd say yes. Why does it need to be done under informed consent? Well, I mean, I think I totally agree with you because- Well, you agree with the implication? It's a, yeah. I mean, I think institutions are, they don't want to take a risk, right? So here's the thing. If you do it under informed consent, it immediately doesn't smack of real world data. And if we're going to implement an impact on our patients, it has to transform to the real world setting at some point. And I would argue that you've made the case and that to have consent in the way, everybody then relates to this as some sort of research that cannot translate then to business as usual at the end and cannot transform to direct care. I would argue the right thing to do is to make the case for this being a direct care approach now, get agreement on that basis with all of your centers, and then you can have a cluster randomized study or whatever, okay, which would be ethical and it can just be done and it's real world. And I think this sequential approach of feeling that we need another randomized control trial and we need consent and the people around us can interpret the need for consent as not- So we did the same thing at, sorry to interrupt, but we did at MGH, we first did a little trial, like exactly what you did, and now we've just adopted it. So we just, we're running with it. It's a quality control project. It's a quality control project. So I agree with you. We don't need a consent. We can be bold and we can do this. We can just roll with it. I have one more question if I may ask you. Sometimes trying to expedite the discharge of a patient becomes the foremost priority. And sometimes you can have patients who have congestive heart failure, they have a lower EF, and all you're doing is cardio wording them, saying it's rate related. We'll figure it out later on. Do you think it's a slippery slope we're on? Agree. Sometimes- Exactly. This push to open up hospital beds because your ERs are full, right, and the hospital resources are limited, so you have the chief nursing officers or the CMOs making rounds in the hospitals and telling the hospitalists and everybody else involved in the patient's care to get them out as soon as possible, right? So as a result of which, there is this inherent rush in people's head and not really optimizing their care. I think they're all missed opportunities, number one, to optimize the patient care. A proper education, right? I think we don't really use that opportunity when they're really sick to drive down the points of why they are sick and what else needs to get done in that. So I think this is where we as sub-specialists, I think we had to own some of those issues and push back, right, to really, for the best outcomes for the patient. Right. Yeah. And I think that's where you create your own institutional protocol based on the, you know, the level of severity of the situation for that patient and don't kind of have a one-size-fit-all strategy. All right, with that, we're going to thank you, Dr. Lacheretti. Thank you so much. If I may call upon the next speaker, we have Dr. Rohit Mehta from Advocate Health, who's going to talk to us about wearables versus ICM for long-term arrhythmia monitoring. Looking forward to the talk. Do you know how to get out of that? You're going to press escape first on the top left and just, yeah. Thank you. You can start. Go ahead. Okay, well, thanks to our esteemed chairs and to HeartRhythm for having me here today. And Jack, since you can't start the timer, I have innumerable amounts of time, so. Technically challenged. So the purpose of the talk today is to talk really about wearables versus ICMs for long-term arrhythmia monitoring. And I think what you'll hear as we go through this is there'll be a number of themes that DJ presented that will actually carry forward from this. Even though we're talking about the specifics of monitoring arrhythmia, really what we're talking about is global care and the stickiness of care to the patients. Sorry about that. So one of my mentors once told me, when you give a talk, you always want to eat dessert first as your meal, so you should do the same thing in your talk. So I'd like to present sort of the conclusions of the talk so as we go along, you can follow along. So from a conclusion perspective, really we bridge the talk as this question that we're going to answer, but really it's a continuum. And so really a tailored approach is really the right answer. Arrhythmia type, symptoms, costs, they'll all play big roles in the right monitor, right patient approach. ICMs and continuous data flow will be important in scientific development, so I think we're at an unprecedented time in data collection where we have to sort of reevaluate what the true course of disease is and take a modern lens to that. The other thing is intermittency of symptoms and arrhythmias will also play a role in that. I'm a big believer in consumer-based healthcare, so I do think that a lot of the digital health tools and wearables will actually empower patients to be purveyors of their own primary care, if you will, and afford them the ability to gain that expertise. Really a lot of this is centered around atrial fibrillation, so you'll see a lot of what we talk about today talking about AFib. Really care today is improved by long-term monitoring with the overlay of compliance, so the ability for a patient to be compliant with the transmission of data, the ability of the patient to, once the novelty of a lot of the wearables has worn off, to continue to utilize that. They'll also play a role in episodic care, so in patients that need antiarrhythmic drug loading, hospitalization prediction, intensification of outpatient care, these digital tools will enhance our ability to do that. We won't touch as much on that during this talk, but I do think it's important to understand that that's evolving pretty quickly. So what is a wearable? We think of this commonly as things like Apple Watches, Fitbits, CardioMobile was really probably the initial foray into this market, and really you think about this in terms of duration from hours to 30 days to beyond. It tends to be best fit for things like palpitations or a-fib screening, so in patients who don't have a diagnosis, who have very intermittent symptoms, this is really a place where a lot of wearables will come into play. The pros of these, they're not invasive, they're accessible. You can get a lot of these on Amazon. Real-time transmission, a lot of closed-loop feedback, so one of the advantages I see in this is the ability for the patient to participate in that diagnostic as well. Cons, they're limited for infrequent or asymptomatic arrhythmias, short battery life at times, oftentimes the need to remove these things during sleep, so you don't actually have that ability to get sleep, and certainly false positives that arise from this as well. From an ICM perspective, we commonly will know these as loop recorders, right? So the Medtronic Link, Abbott Confirm, Lux, Biomonitor. Generally, battery life is the constraint here, so about three to four years of continuous monitoring, so the advantage of this is that it is continuous, and the data transmission becomes an important piece. It's really, really good for syncope, right? So the patient can't participate because they've passed out, the ability to capture arrhythmias, but also on the negative predictive value. So in this situation, the ability to rule out a cardiac cause for syncope, I think, becomes important. We all know the data around cryptogenic stroke, and for rare arrhythmias as well. There tends to be a very high diagnostic yield, high fidelity EKG recording, and then the remote monitoring piece of this. The cons, it tends to be still minimally invasive, but it is invasive, and a higher cost. The other thing is the follow-up infrastructure becomes important. So the provision of the monitor is one thing, the provision of the information is another, and so the ability for a patient to comply with that remote monitoring piece becomes important. Understanding data limitations of the device as well, so if a patient's not by their monitor and they've had multiple arrhythmias, the ability for some of those arrhythmias to be wiped out and lost. And this is sort of an oversight in terms of a lot of the trials that have been done in AFib over the past two years, or really year and a half. And a lot of it is geared around the initiation of anticoagulation for relatively lower risk patients. What do we do from the standpoint of the congestive heart failure patient, anticipation of hospitalization. We talked a lot about resource provision with regards to patients and hospitals, and how do we manage those data pieces. So the big piece of data I want to really revolve around was actually one that DJ's group that had led. So it was called Monitor AF, it was a retrospective multi-center study of ICMs versus routine EKGs and long-term monitoring. So this included holters, events, whether that be, they didn't necessarily control for 48 hours versus two weeks versus a month. And they split the patients into two big groups, so about 1,000 patients in each group, fairly well-balanced overall. With ICM, there were increasing attempts at the provision of rhythm control in these patients with ICMs. So with anti-arrhythmic drugs, there was a higher provision of anti-arrhythmic drugs, higher incidence of catheter ablation in these patients as well. So ICMs tended to be more favorable on a number of things. Post-ablation and post-anti-arrhythmic drug, there was a higher freedom from atrial fibrillation over the past, over the 12 months. Freedom from anti-arrhythmic drugs post-ablation was much higher in this group as well. Anticoagulation, higher degrees of anticoagulation penetration when appropriate, and reduced risks of stroke, likely from the anticoagulation provision. Again, reduced risks of major bleeding. So what I take from this isn't so much that the monitor itself is magic, but it ties the patient to us. So it ties the patient to electrophysiologic care, the ability for that patient to come back on repeated episodes. And once you're past the 12-month surveillance period, patients that have late recurrences are tied back to their electrophysiologist versus to other parts of the healthcare system where access may be more at a premium. On the wearable side, really, the initial pieces of wearable data, probably the groundbreaking pieces came from CARDIA. But probably the big study was really Apple Heart looking at the Apple Watch. And there were some interesting pieces from that. Number one, patients enrolled in a clinical trial at unprecedented levels. And it was probably one of the shortest times for enrollment for the magnitude of the study. But once you took a patient from that to intentional treatment, meaning that the provision of a monitor that they had to mail back, compliance rates really dropped off. The provision of a virtual visit, compliance rates further dropped off. So it really illustrated that with regards to a lot of the wearable data that we have, the patient has to be an active participant no matter how easy you make it. So I think that's a big, big take home from that. And when you look at this, overall, wearables are good for short-term arrhythmias in symptomatic or at-risk patients. ICMs really tend to be the gold standard for rare or silent arrhythmias and stroke prevention. But they do have complementary roles. And there are patients that we do have. So one of the challenges in the ICM population is the patient doesn't have direct access to the data. So if a patient has palpitations, there's potentially a burden on the healthcare system with phone calls, messages, and things that come into our device clinics to anticipate or articulate what that specific data piece was. So we do have patients that actually will employ both. They'll actually have a Cartier or an Apple Watch concomitantly with their ILR. And it helps kind of bridge and balance the two things. The other barriers from a wearable perspective are technology gaps, so the ability to use the technology. And it's interesting because we don't see it in the places where we think it would be natural to see. A lot of our older patients are actually pretty facile at using the technology and embrace it at a much higher clip. And really, it's sort of in the middle age, kind of 40 to 60, where we're starting to see drop-offs within our population because of work schedules, life schedules, those sorts of things. And they don't actually employ that as much. Closed loops of care. So one of the things that we've encountered is large volumes of data that are coming our way sort of interpolated into the normal day-to-day practice of medicine. So you're going through a normal clinic day, a normal lab day, and then you have this huge volume of data that you have to analyze and assess and complete with phone calls, messages, responses, and the things that go along with that. So there's no immediate give back to the patient. So one of the things that I always think about is a device that used to be in place in heart failure called Savicor, which is a left atrial pressure sensor. Patients at that time would put a device over their, at the time it was a PalmPilot, over their device, they would give them a prescription of what to take. So that closed loop piece, I think, would be important to slowly develop so patients have some sort of a prescriptive response to the data that they do see. Data management and facilitation, this is large, large swaths of data. And then reliability of data. There's still a fairly hefty false positive rate in this cohort as well, especially on the PPG side. So conclusions again, to go back over this. Arrhythmia type, symptoms and costs are gonna be big, big pieces to this. I do think the wearables do help defray the costs. So they tend to be much easier for the patients to get. Compliance is another piece. Data compliance is another thing that we need to talk about. And then in general, what I would say, wearables are generally better for screening. So generalize your symptomatic screening for patients who have palpitations, for patients who may have had a singular episode of atrial fib, looking at overall burden in the symptomatic case. ICM's tend to be superior for long-term monitoring and healthcare adherence. So they tend to bring patients back to within our purview and our longitudinal long-term care. Thank you, I'll take any questions. So mics are open, any questions while we wait? Can I take the liberty of asking a question? Of course. So this whole area of AIECGs where from a simple 12-lead or a single-lead, you can actually predict with a fair amount of success, depending on what you're looking at, whether it's AFib or VT or a PAWS or a systole. And even from patch monitors, which are not conventional wearables, but patch monitors for a 14-day from the first few hours, you can predict what's gonna happen in the next 14 days. Do you think some of those algorithmic tools are gonna cannibalize the wearable and the ICM market in the future? And what are your thoughts of how that potentially could evolve? Yeah, I think that's a great point. Obviously, it's a little daunting coming from one of the experts in this space. But what I would say is, I do think they will cannibalize some of the more rudimentary wearables that are there to date. To me, AI has two major things that hopefully hold promise. One, efficiencies of care. And I think on a day-to-day basis, the ability to really compartmentalize information into discrete nuggets that we can handle on a more day-to-day basis. And then the other thing is risk prediction. I think that is a big piece of it. The one place where I still think I have some concerns is really what the give back to the patient is. So the ability for that patient to be able to see their data and understand what's happening to them in real time and potentially be a more active participant. I tend to have a few outlier ideas with regards to healthcare. I'm a big fan of saying the patient should own data. They should be able to provide that data, understand that data, have tools that they can help them conceptualize that. And I do think that that will help both the cost and the provision and access to care. So I do think the AI pieces will help a lot with that. Great, Dr. Germany. That was a great lead in. So Robin Germany from Minnesota. So in diabetes, they've been doing this for a long time. There's this great feedback. I remember, because I'm getting older now, that people were a little nervous that people would overtake their diabetes medication, all these issues. So finding discrete places in the cardiac space where our patients can monitor their own. So I'd be interested to hear as you look through all this data, where are areas that we think that perhaps patients could take control and be able to do those things without perhaps causing damage, right? We don't want them to change their sodal dose or something like that. I think that's a great question. I think we tend to paternalize our patients a lot. And I do think our patients are a lot more sophisticated than we give them credit. I think in the heart failure space, I think the Savicor device did teach us that, that they did have the ability. And patients sort of understand what things they need to take at what times, and they sort of develop that expertise with their own care. They understand when they ate inappropriately or did things inappropriately, and they can basically overcome some of that without a call to the office. You know, it's interesting you mentioned the diabetes space. There's actually an app called GluRU that it's actually a college classmate of mine that he developed. And because he's a computer science programmer, he could really rapidly iterate it. And so what happened was he developed this for his son, who's a type one diabetic, because his son would constantly leave his phone in the car. And it was the ability to distribute that information not only to his son, but to the entire care team. So if his son left the phone in the car, he would get an alert. In fact, the alert was so delayed initially that within three days, he could figure out what the timing of that would be. But it basically empowered his son and illustrated to his son the importance of this. And soon he didn't really have to really oversee a lot of that. So I think in cardiology, like I said, I think the question is fantastic because I do think patients can manage a lot of this on their own. There may be a 10 to 15% swath of folks that we need to spend a little bit more time with. But then we now have the access to be able to do that. You see any questions? I do actually, just to bend your mind a little bit more. Two things, that one, you said you believe in the consumer technology, that the patient should be responsible for the cost of those, I think that's what you said. And secondly, the closed loop communication and how important that has seemed to be with the previous devices. With the CardiMobile, I believe that you can subscribe to or pay for kind of a closed loop system like that where your ECG gets loaded up, somebody interprets the data and sends it back to you. But there isn't necessarily the next piece of an EP or your own personal care team taking care of it. So what would be, I guess, combining those two things in an ideal situation to solve that gap? Yeah, so I think a couple things. When I say they should be responsible for the cost of their care, number one, I think our obligation is to make sure the cost of that care is sufficiently low to allow that provision. So I always tell people all the time, one of my big issues with healthcare is it's the only place where effectively you go to a restaurant, right, your doctor, you get assigned a meal, your care, and then you go home and the bill is provided to you later. So no one would ever go to that restaurant, right? But yet, that's how we continue to treat care. The ability to go to Amazon and order the cardio device, right, and when Dave Albert developed it, right, the initial price was I think $120, $130. You've seen cost attrition on that front, and I agree with the subscription service, but there also is the ability to just do a one-time only recording, which then gives you the immediate interpretation that has not been verified, but it's pretty good. And ownership, right? And maybe that wipes away and you can't store them historically, but you can continue to download. So it may be a higher hill to climb, but it is definitely a hill that you can still climb. So I do think those things are gonna be important. I think a healthcare marketplace, if you go to other countries, in Nepal, you can walk into an imaging center and order a cardiac MRI for yourself and pay out of pocket because the cost is achievable. So I do think those things are important, but here I do think there has to be the provision of the appropriate education to take that information that patients can use. My mom's 83, she can use YouTube to almost solve any problem in her house. Our patients can do the same. I completely agree, and I think the only way to make healthcare sustainable is for patients to have skin in the game. And I think the whole process also needs to incentivize patients, right? And I think, for example, your healthcare premium will be lower if you maintain your own hemoglobin A1C less than 7.0 as opposed to it being above 8.0, whatever, a cutoff that is created artificially, but can be, again, individualized and create better habits, wellness, through these monitoring strategies. Last question to you, though. Wearables and implantables are still expensive. And there is this whole concept of digital divide. Do you think we should be spending more time using machine learning strategies of EHRs to risk stratify patients initially for those maybe at high risk, moderate risk, low risk, and then decide how to then deploy these wearables and whether these wearables should be a part of the treatment strategy so there is no cost to the patient for it? Thoughts? Yeah, no, I think the healthcare economics could potentially fund this, right? And I think that's the question at hand. So if we were able to save three or four hospital days, that you could really look at the downstream cost of that as the ability to perform that, right? So one of the things that I think is gonna be interesting over time, our health system, our CEO is on the board of Best Buy, and Best Buy has the Geek Squad, and they're actually looking at transitioning that to more of a healthcare delivery piece. So a distributed model, everybody has a Best Buy in their home state and town, and so the ability for us to employ health tech through that model rather than fixing a CD player or other things, they're actually going to the home, taking vitals or potentially troubleshooting the wearables or the health tech that they do have. So I do think those models, and certainly the funding through that, I think is a great option. I still am a big believer of bringing the costs down. I think worldwide, we've proven that the costs in the US are significantly higher than what they need to be, and cutting out the middleman, like DJ said, I think we can really get to that true cost of care or close to the true cost of care, and patients can afford that. Great. Thank you very much. Thanks. Thank you. All right, and next I'd like to invite up Dr. Bren Summers, speaking on sleep apnea and AF. Can digital health help decrease AF? Let's see now. How do I get out of this? Escape. Right, left, yeah. Ah. Oh, I see, I see, great. So thanks Lisa and Jack for the invitation to speak here today and the other speakers set a great precedent for some of the stuff I'll share with you. I'm gonna be talking about sleep apnea and atrial fibrillation. Can digital health help decrease atrial fibrillation? Forward this one right here. Okay, let's try. You can use the keyboard or you can use the other. Okay, we'll just use the keyboard. Okay, these are my disclosures and the reason I've got the red is because those are the companies whose work is actually very closely tied to what I'm going to talk about today. That'll be iRhythm, Sleep Number, and Zoe. I'll be talking about four things. First, why do we care about sleep apnea and atrial fibrillation? What are the reasons they're linked? Second, how do we monitor sleep? Because in order to understand picking up atrial fibrillation in sleep, we have to pick up how to monitor sleep. Then what are the options for simultaneous sleep, ECG monitoring, and then we'll summarize with the take home messages. Sleep apnea and atrial fibrillation, to understand that, let's think of what happens with normal sleep. Normal sleep, the recordings you see on the top, these spikes, these are microneurographic recordings of sympathetic neural outflow from the brain to the peripheral blood vessels. In a sense, a window on the sympathetic system to the blood vessels. And this slide shows that when you go from wakefulness to deepening stages of normal sleep, you get progressive decreases in sympathetic activity so that when you're in stage four, your sympathetic activity is very low, heart rate is slower, hour interval is longer, and blood pressure is lower. In REM, which is the time that you're dreaming, sympathetic activity is very high. And heart rate and blood pressure are extremely labile. You can see wanky bark, you can see long pauses. It's purely physiologic. So keep that in mind as we try to differentiate between this normal sleep picture and somebody with sleep apnea. This is a newly diagnosed, never treated, otherwise healthy patient with sleep apnea. This top panel shows sympathetic activity, breathing, and intra-arterial blood pressure when he's wide awake. Pneumoxic, everything's fine. Blood pressure is 130, 80. He's got high sympathetic activity. We know that's true for apneic patients. Then he goes to sleep. He's in REM sleep. REM is the time apneas are worst because you lose your postural muscle tone during REM, which is when you're dreaming, so you don't act out your dreams, but you also lose tone in the upper airway, so the tongue falls backwards. The airway is much more easily collapsed. And so whenever you're in REM, that's when your obstructive apneas are worst. Blood pressure and sympathetic activity are very, very high. This is because your hypoxia and hypercapnia stimulate the chemoreflexes. High sympathetic drives, vasoconstrict, blood pressure goes up to 240 over 130. From a baseline of 130, 80, you treat the patient with CPAP and blood pressure is much better, sympathic activity is down. So here we have to figure out if we want to recognize sleep apnea from non-invasive recordings, can we pick up these kinds of things? The blood pressure increases. The RR changes. You see how the RR increases towards the end of apnea? That's purely physiologic. That's a diving reflex. On an ECG recording, so we pick up this patient probably has apnea. And also, can we pick up these sympathetic vasoconstrictor signals? And I'll talk about the watchpad ZOHO device which can actually do that. So we think of a-fib and obstructive sleep apnea. Why are they linked so closely? Because sleep apnea causes hypoxemia. It causes sympathetic and simultaneous vagal activation. It causes extreme blood pressure surges which you saw. You didn't see transmural pressure gradients because when you're obstructed, you generate this negative pressure of minus four in the chest repetitively and that eventually stretches the atrium out which you'll see shortly. And it also causes systemic inflammation because of repetitive hypoxemia, reperfusion, all of those from sleep apnea, exactly the things that cause atrial fibrillation. And so that's why we see this high likelihood of sleep apnea and a-fib existing together. So what happens to the atrium in people with sleep apnea? This is the patient with sleep apnea in the top panel here. And this bottom panel is a normal control. It's obvious that the sleep apnea patient has bigger atria because they're constantly being stretched every night, throughout the night, for years. And along with the stretch, you get scarring, you get double potentials, fractionated potentials, creating the substrate for developing atrial fibrillation, particularly when you throw in hypoxemia, hypercapnia, and respiratory acidosis, et cetera. So the evidence that sleep apnea does in fact lead to atrial fibrillation in the long term, these are about 3,000 patients seen at the Mayo Sleep Clinic. And we tracked them for about 15 years to see who develops atrial fibrillation and how it related to sleep apnea. And it turned out that if you had no sleep apnea, this yellow line, you really got about a two to 3% likelihood of atrial fibrillation in the long term. If you had severe sleep apnea, the top line there, you had a 20% chance, about 10 times the likelihood of someone who didn't have sleep apnea. That's one message. The other message from this slide is this is a moving target. So you could look at a severe sleep apnea patient today and look for AFib, you don't find it. That doesn't mean he's not gonna get it in the future. So keep that in mind with your severe sleep apnea patients that don't have AFib now, there's a one in five chance that it's gonna come sometime in the future. So that's something to watch out for. Now I only have this slide because it's interesting, not much to do with what we're talking about, but this is a patient who came in incidentally having a carpal tunnel release, obese patient, had some hypoxemia, we studied her in the sleep lab, she had atrial fibrillation and this, by the way, we do what we call split night studies where we will study them for the half night, diagnose the apnea, treat them in the second half so we save a sleep study, we save them coming in another night. And so what this patient had was bad sleep apnea, this is the hypoxia or the oxygen saturation. She's very low, right? This is the heart rate, it's all over the place because patient's in AFib. And then my colleague, Sean Cables, put a CPAP mask on her and look what happened, the oxygen normalized, but look at the heart rate. She went back into sinus rhythm. It was maybe purely coincidental, but it's interesting enough that I thought I'd just show that in terms of the effects of acute apneas possibly on developing atrial fibrillation. So we've seen why sleep apnea can cause atrial fibrillation. Now let's talk about how we monitor sleep. What are the standard ways of monitoring sleep that you have to think of if you're gonna develop a monitor that's gonna pick up sleep apnea and atrial fibrillation? The first is a type one. That's the regular sleep study that we all know and hate because you come in, lots of bells and whistles, you have to stay overnight, it's attended, that's a type one. Type two is the same thing, but not attended. You can do that at home, you can do it in the hospital. Type three is what we know of as home sleep apnea testing. That's where you have about four to six cardiorespiratory channels and it's actually accepted for Medicare billing to do that. And type four, that's type three. Type four is overnight oximetry, something that I actually like a lot. And that just is an oximetry measure with potentially a heart rate measure as well. What do you see with overnight oximetry? This is what you see, this is the oxygen saturation up here and you can see a couple of things here. One is it's clustering, that's what happens. As soon as you get REM, you get your bad apneas and REM is not the whole night, so that's why on oximetry you'll see clusters of desaturation. Those desaturations will get worse towards the end of the sleep period because that's when REM is most likely to occur because that's the circadian pattern of REM sleep generation. It's most marked at the end of sleep and this is the reason why if you're using CPAP but only four hours a night and you take it off in the first half of the night, right, you take it off there, you're left unprotected for the worst apneas that come later in the night and you didn't really develop that acclimation to hypoxemia that could potentially protect you from something bad happening. So that's oximetry. Now, what about simultaneous measurements of sleep apnea and the ECG, which is what we're here to learn about and what the options are. But before that, we have to recognize a couple of things. Oh, by the way, all of these solutions I'm talking about are, again, moving targets because they could change tomorrow. They may have changed last week and I don't know about it. So if my information is outdated, I'm sorry, but it's just the way this field is developing. But before we talk about that, these data from Prosh Sanders in the VARIOSA study and what you see are three patients. On the left, mild apnea, middle, moderate apnea. On the right, bad apnea. And those lines, those colored lines represent the severity of sleep apnea on each day in patient one, patient two, patient three. The message for us on patient one is that even though this patient we think has mild apnea, look at the reds. It just depends. This is a movie. You take a photograph of one night, that really does not represent what's going on for the whole year. This is why wearables or remote monitoring are so important in sleep apnea because it changes from day to day and a mild apnea can have really bad apneas on some nights. A severe apnea, right, can have actually minimal apnea on some nights. Just depends when you catch them, what they did, how much alcohol they drank, how much sleep deprivation they had. So keep that in mind, that the one number we get that we all hang our hats on is really just a guess because it doesn't really represent what's happening for a constellation of nights. Now, how are we gonna measure sleep and AFib? Well, the Apple Watch, I'm not gonna belabor this. Most of you know that. This shows us how we set it up. Now, the point of the Apple Watch, though, in terms of picking up atrial fibrillation, you guys may have different data, Rohit, from all the different studies that are done. This is just one study making the point that if you put an Apple Watch on somebody, and you mentioned if your arrhythmias, the type of arrhythmia affects the pickup, well, it's also type of baseline ECG. If you have an IVCD, if you have heart block, if you have a pacemaker, if you have multiple PVCs or PACs, it affects the ability of the Apple Watch to pick up AFib, at least according to this study, because of these 700 patients, specificity in these people with a baseline abnormal ECG was only 69%, was 81% with sensitivity, 69%. Now remember, when Apple Watch picks up AFib that's not really there, the false positives, that generates a whole host of downstream costs that we have to follow because we have to, this patient may have AFib, we've got to either look at the original recording that triggered that call or we have to do a whole bunch of tests to rule out the possibility of atrial fibrillation. But there may be other solutions on the horizon. This is applying a AI algorithm to help the Apple Watch make up its mind in terms of whether it's really AFib or not, you can see in terms, look at the bar on the right, actually, the two bars on the right. Sensitivity and specificity. Apple Watch 42 in terms of sensitivity, deep neural network is 96. Specificity, Apple Watch 95, doesn't change much, but when you add AI algorithms to help process the Apple Watch signal, you can actually boost your ability to pick up true AFib or exclude true absence of atrial fibrillation, we think. Now, Apple Watch picks up AFib, we think it works pretty good, what about picking up sleep apnea? That's what we want, right? We want to pick up the AFib or we want to pick up the sleep apnea and be able to detect both of them. Well, this is the Apple Watch oximeter or oximetry testing or oxygen detection, which is not approved for medical uses, but also, sadly, or not sadly, it depends who holds it as a patent, because of a potential patent infringement in the US, Apple Watches are not available now for oxygen saturation monitoring, so keep that in mind. Give you the ECG, it won't be able to give you the, at least the newer ones, won't be able to. So, what else do we have in terms of picking up AFib and sleep apnea in terms of wearables? The AuraRing, it uses photoplethysmography to measure heart rate and oxygen saturation. It can alert you to the possibility of sleep apnea being present, but it's not a definitive diagnosis by any means, in fact, all of the wearables, I think you have to take with a little bit of salt until we have really, really good technology, perhaps combined with deep neural networks to help us refine the diagnosis or the exclusion. Now, here's something, comparing a study, comparing the AuraRing to the Apple Watch. One study, I'm not gonna sign on to this or say it's not true, I'm just showing you the data that's available in the literature. First, the AuraRing on the top, for atrial fibrillation, they found 98% sensitivity, 100% specificity. For the Apple Watch, they found 74% sensitivity, 100% specificity. This was done in people after cardioversion. Before cardioversion, they took segments. After cardioversion, they took segments. They said, well, can you pick up what is atrial fibrillation and what is not? So we've seen these wearables. Now, how about the real, honest to goodness, bona fide ECG monitors? There's many of them available. I wanna talk about the BodyGuardian, only because, two reasons. One is I have affection for it, because we helped develop it with preventers at Mayo many years ago before it got bought by Boston Scientific, but also before it got bought, and in that period, a few years, we're trying to develop a sleep apnea, sleep and sleep apnea monitoring system to couple with the BodyGuardian, so we could pick up sleep and tie it in with various arrhythmias, but we didn't quite get that chance, and I don't think BodyGuardian has that yet. I don't have much to do with them at all right now, but it seems that that would be a great addition to that technology. So, there we had an ECG device aspiring to detect sleep apnea. How about a sleep apnea detection device aspiring to detect AFib? This is the watch pad, and the watch pad picks up peripheral arteriometry. Remember the sympathetic activation I showed you earlier with the sleep apnea? Well, the watch pad is there to try and pick that up peripherally, and so it tries to diagnose apnea using that and oxygen saturation, body position, heart rate, actigraphy, snoring, and so what the watch pad people have done, what Zol has done, who gives us equipment, by the way, I should clarify, they give us watch pads for some of our studies. What Zol has done is developed an algorithm to use that tonometry and heart rate, or heart rate data from the photoplethysmography from the oxygen saturation monitor to try and predict whether or not someone develops atrial fibrillation, and there they found a sensitivity of, on the green at the bottom, sensitivity of 77% and specificity of 99% in terms of detecting AFib. So, we've seen many things. We've seen ECG monitors trying to detect sleep apnea, sleep apnea monitors try to detect the ECG. How about a bed, okay, a bed where you sleep in that can pick, that tries, that tries to pick up sleep apnea and atrial fibrillation? Here again, I serve on the scientific advisory board. They also give Mayo money for research, and so very quickly, the bed has these sensors that are ballistocardiographic, and you see on the top here a sample ballistocardiograph. Now, you break that signal up into components. You find you can pick up breathing. You can pick up the heart rate, and from the heart rate, you can pick up the heartbeats, okay? The cardiac component is there, and this is the heartbeats, and it tries to use that combination of signals when you're sleeping to pick up whether you have apnea and whether you have atrial fibrillation. How does that work? To pick up atrial fibrillation, so far the algorithms show that they have moderate sensitivity, 62 to 66%, and specificity, 59 to 63%. To pick up sleep apnea, a little better. Sensitivity is 71%, and specificity is, I think it says, 83%, okay? So here's something that's just coming out of the blue, a bed trying to do both measurements on a regular nightly basis. So what I've tried to do is talk about why sleep apnea causes AFib, how we monitor sleep, how we can take heart rate monitors, make them pick up sleep apnea, how we can take sleep monitors, make them pick up AFib, and maybe a bed, maybe other things you sleep on that you use every night can do that, but overall, I mean, the key thing is, you really need to do this on a nightly basis, and then even if this year there's no sleep apnea or no AFib, it doesn't mean it's not gonna happen down the road, so use the clinical judgment to decide who gets it. Thanks very much. We're open for questions. We can take a couple of questions. We're over the time limit. Lisa, you wanna go ahead? There was a great talk, thank you. Lots of new knowledge happening here in this brain. One product I know that is on the market, the Withings Sleep Mat, but I believe that that's just for apnea and disordered breathing. There's no ECG component, or? No ECG, you're correct. The Withings picks up apnea. You stick it under the mattress, and it purports to, it's good, it's okay, actually. It's good at doing that, picking up apnea. Are there any of these direct-to-consumer devices that can be used to diagnose sleep apnea? Direct-to-consumer, I have no recommendation. I'm so sorry, I have no recommendations on that, unfortunately. No problem. Sir, go ahead. It's a really interesting talk, thank you. Thank you. So I was interested in the variability in obstructive sleep apnea. Does that correlate at all with episodes of paroxysmal AF? Great question. So that was done with a pacemaker, which could pick up the apneas, right? And what they found, what Sanders Group found, was actually, when the apnea was really bad, that's when your AFib burden was most marked, and exactly what you would intuitively predict. But more interesting was that the AFib actually, when you look to the phase delays between the apneas and the AFibs, the apneas preceded the AFib. Because there's this thinking that maybe AFib, because it's mucking up your heart, your cardiac output, may be triggering apneas, because that's what heart failure does sometimes. But it actually, the sleep apnea precedes the AFib, talking about a causal relationship. So as a sleep apnea expert, should we be looking for sleep apnea in all patients with paroxysmal AF? I don't know about the first part of your question, but the second part I can tell you that probably not, because we don't have the money to do that right now, until we have a really cheap way to find it. But look for the ones that are refractory to treatment, the ones who are obese, the ones, obviously, who have daytime sleepiness. Those are the key things that'll tell you, especially the big bellies, are the sleep apnea markers. Go ahead, sir. Please introduce yourself before you. Excellent talk, Viren. I'm Rajiv Mahajan from Adelaide, South Australia. We've been using, I was on one of the, on the paper, the various study. We've been using the weaving mat quite a bit in the studies, and like all these digital devices do have limitations. So the weaving mat has a limitation that the sleep block of four hours before it can diagnose sleep apnea, the EHR index that it gives. With the other variables that are there, what is the limitation? How much, what is the length of the sleep time which is required for it to actually give some reliable data? Usually, most of the ones take it for the whole, or you're saying, what is the minimal sleep time where you can make a call on the apnea? Hell, that's a hard, it just depends on the patient, the pattern of the apneas. We don't make a call on sleep apnea on anything less than three hours of sleep. So, because remember, very quickly, apnea depends a lot, obstructive apnea, depends on REM sleep. REM sleep doesn't come on straight away at night. It warms up, and the longer you sleep, the more frequent REM becomes, the longer REM lasts. So, if you just got two hours of sleep in the beginning, you don't have much REM, you're not gonna have much obstructive apnea. So, is there, for example, with the EHR that we get with the variables, what is the, they do give the EHR 15 or 16. Yeah. How does it work? So, is it slots of hours? So, when EHR per hour, so how does it work? Because for them, is there a block that they require to actually calculate this? I don't know. I don't know how each one's algorithm works, but generally, the problem with the variables is they can't tell you when, they'll tell you when someone is lying in the bed, right? And often, they will measure sleep from the time you lie in the bed to the time you get out of the bed, but you weren't sleeping all that time, but it divides the number of apneas by their total sleep time as opposed, assume total sleep time as opposed to your true total sleep time, which is the time you really were asleep. That's how you should. So, generally, you can guess that it probably underestimates the AHI because it overestimates the sleep duration. Thank you. Sure. Thanks. Great talk. We're in. On that note, I would call this session to an end. I want to thank my co-chair as well as all the faculty for phenomenal talks and all of you for taking the time to actually be here. Thanks again. Thank you.

digital health

electrophysiology

atrial fibrillation

sleep apnea

wearables

implantable cardiac monitors

patient care

early detection

health monitoring

electronic medical records