There are a lot of issues surrounding the field of VT and sudden death prevention. And let me first ask Andrea, you presented a very clean case, the kind of case we love, you know, the kind of case we save tracings from and show at conferences and stuff, which is why you had all those nice slides. But in the real world, we know they sometimes are not that tough. When you have a really challenging case where you're on your eighth VT morphology, when you have the patient who is hemodynamically unstable in the lab, how hard do you push? What's your threshold? What's your decision process in terms of going after a complex substrate in the lab? Yeah, that's a great question, and that's the more typical case probably nowadays that we see. So it really does depend on the particular patient. So the one key primary factor is patient safety, right? We want to make sure that we're not doing things to the patient, that they don't have prolonged hypotension. And sometimes, actually, after they've been having all these arrhythmias and they're getting drugs for sedation, their blood pressure may be low even when they're in sinus rhythm. So I think as much as we go for as many VTs as you can ablate, we know that sometimes, hey, listen, it's patient safety. We did what we could. Maybe we took the most common clinical VT and got rid of that now, but the patient has some risk here. He's unstable. He or she is unstable. They're getting more and more volume during the procedure, and so we stop. So you need to know as much as how long to persist is when to stop and say come back another day, whether or not patients need some kind of blood pressure assistance with devices. But sometimes you do just say, I'm going to do what we can today. And then sometimes you'll see there might be an epicardial VT that you might, depending on what center you're at, you may not be prepared and set up to do an epicardial ablation on top of endocardial. You just have to come back more than once. And I think that's completely acceptable. The one thing you don't want to do is cause other organ damage and compromise to the patient. So if there was a more clean case. I couldn't agree more. This is sage advice for any ablation group. We also have published data that shows that the longer you keep a patient on the table, it actually has an adverse outcome. This is published data sets. Therefore, you can use a John Wooden line, be quick, but don't hurry. So I think that is our motto for catheter ablation of VT. And getting a lot of information ahead of time greatly improves outcomes so that you can plan the procedure better. And it's also a situation where you need to have very good partnership with your heart failure physicians so that the patient is as optimized as possible when you bring them into the lab. There's a little bit of misconception with industry to think that, ah, we can just put in a support device and that will take care of all the problems. Hardly true. You're not just dealing with one little number during a procedure. It's everything that happens before, everything that happens during, and everything that is going to happen after the procedure. So you have to really look at it as a team of teams. And the link between your CCU, your heart failure doctors, your anesthesiologist during the procedure, and again, the CCU, the heart failure doctors post-procedure for some of these very sick structural heart VT patients is the difference between life and death. And it's perfectly fine to come back, live to fight another day, because this patient has had this problem which has built up over 15, 20 years. You don't have to fix it in three to four hours. So, Dr. Kim, are you an early ablator or a late ablator? Do you wait for the patient to fail high-dose Amio, or do you look for that case straight out of the gates? I mean, I think it depends on the circumstances for sure. I think the older patients who are post-MI, who you don't envision on Amiodarone for decades, I think it's fair to give them a trial of Amiodarone. But for younger patients who might come in, I think our threshold is lower to bring them to the lab. But I would fully agree that it's sometimes hard from an ego standpoint to be in the heat of the battle and sort of admit, I guess, short-term defeat. But it's definitely important to be judicious about when to stop. And Dr. Passman? I would agree with Susan. I have a pretty high threshold for taking the average person because things can go terribly wrong. I mean, most of what we do in EP, if you're sticking sort of the upper chambers, are quality-of-life issues, right? We're not talking about life and death. But this is definitely a situation where you can make people worse, right? You leave them in VT for a while. You wind up putting them into VF, shocking them multiple times. And this downward spiral can initiate where, if you don't know when to stop and say enough is enough, I'd rather be semi-successful and bring you back, you could do real harm in trying to help someone. So this is a situation where, truthfully, I don't think that you should, you know, if you're in a center or you have a patient in a center that's doing five of these a year, that's probably not a great situation. This may be one of those situations where you transfer that patient to a place that really does a lot of this because things can go terribly wrong. This is phenomenal advice. And it's not an ego thing about how many cases you do. Even with centers that do 200 of these a year, we are the ones who actually say stop right now. And there have been some instances where within one hour into a VT ablation, and some of you may have picked on the subtle point that Dr. Russo made, the patient may be in sinus rhythm, but a little bit of sedation actually causes hemodynamic instability. So that tells you that many patients are walking around somewhat pseudo-stable. We use that term a lot on clinical rounds in medicine. So pseudo-stable is people who are walking on thin ice. So, again, it's judgment. And judgment, of course, comes with experience. And the worst thing that you can have is your ego get in the way. So that can clearly be stated. We want to be old pilots because there are no old, bold pilots. So of the panel, who uses hemodynamic support on a regular basis in their patients? We use it regularly in appropriate cases. So it ends up being about one to two cases a week. And these are extremely well-selected patients in whom there's no option but to use elective hemodynamic support. And I'm not talking about patients who already are in VT storm, who already have an assist device that has been placed for some reason, an impeller, whether they came in with an acute event. Electively, the reason to do that is very, very defined, and it's a very small subset of patients. But we do think that this is one area where we really wanted to, you know, we were hoping that this would dramatically create exciting new indications and outcomes for patients. But the jury is still out because no one can agree. And it's very hard to define the classic patient who must need the support. Maybe Andrea and others can comment on that. No, that I agree with. And I think it also depends, and you alluded to, it depends on where you're at and what center you're at. So I had worked at Penn for 12 years, and you have this whole infrastructure of this heart failure team and everything else going on. So I would say, and I now, you know, pre-select patients a lot more and have no hesitancy whatsoever to just call my former colleagues and transfer the patient over. If I know that this is a super sick patient, you know, I would send him out to a center where I think that patient would be better off. So I'm not, you know, routinely using it. I would tend to do, if you want to consider the more stable patient. So I think, you know, the real shift in the last decade has really been the experience with, excuse me, substrate mapping and ablation as opposed to, you know, in the old days we used to have to start the VT and have to map the circuit, and we were aiming to kill off that critical isthmus that the VT depended upon. And as Rod was saying, these people start out looking great, and then they're on that slow spiral. And, you know, by a couple, three hours into the case, they're hypotensive and their VTs are faster and polymorphic, and it's a real mess. And my personal approach, and I'd be interested in the panel, we do relatively limited VT stim. Relatively, we like to do an induction to get the morphologies so that we have the morphology in the lab that we can compare pace mapping sites from, but most of the mapping is voltage mapping, pace mapping, and looking for these high frequency late potentials that correlate to the substrate of slow conduction zones in the border zone and through the body of the scar. What does the panel do? I think your point is well taken about the value of substrate mapping, and so oftentimes we'll, you know, our first step is to map the substrate and understand where is the scar, where are late potentials, and then once we kind of have our template down, then, you know, perform an induction and kind of target the areas that we think, you know, are going to be fruitful or high yield in terms of ultimate ablation. Yeah, well, I'd love to get the inside of the panel. I mean, I think it's somewhat naive of us. I mean, we all have a slide that shows, you know, I terminated this VT in one burn, but this is not a unidimensional problem, right? This is transmural. You know, this is a diffuse process, and are we thinking about this properly? I mean, Shiv, what you showed that the heart's only, you know, a third myocytes. Well, I actually didn't know that, but boy, we spent a lot of time thinking about that one third, and again, maybe conceptually we're not viewing this appropriately as a disease that's focal that we can treat with a discrete number of lesions. This is, I love the comment that Dr. Passman just made, because this is what you don't see in the literature, all right? Human beings are human beings. We always like, in any published paper, people put out the best image that they have, and you always can sort of stage it, not in a bad way, because that's actually what happened to the patient. The outcome was good, but everything that happened before, and we try to capture this in live cases, but the point to be made over here is be very smart about how you approach. One size does not fit all, and I completely concur with the point that Dr. Haynes made, and that is unnecessary pacing is perhaps the worst iatrogenic thing that you could do to the patient, because it's better to define the substrate and be very targeted about inductions, and for some of the trainees in the audience, if any of the fellows showed up, they should go look up the citation classic paper that Fred Moradi published from University of Michigan. This is a non-ablation paper where he repeatedly measured inpatients. We do this all the time in the EP lab, right? S1, S2 stem, and after eight S1 stems, he just collected coronary sinus blood, and he showed that the lactic acid levels already went up. So when you're pacing the heart, and the heart loves to be paced and activated through its own wiring system, so if you pace it for an abnormal location, it's actually not metabolically good, and already these patients are quite sick. So that's kind of how I look at the big picture, because the two-thirds of cells in the heart are also metabolizing, burning glucose, doing a lot of other things. So you want to keep the entire environment happy, and sometimes that is inflamed. You know, we don't pay much attention to many of the other processes that are going on. So I think that's where optimization, being smart about understanding substrate, if you have wideband MRI, use your radiologist. They can give you very useful information, and you can focus your mapping to a given area, and that's when the targeted ablations occur. And I think this late potential ablation has helped the field because you can get away with less extensive ablation compared to what we did 15, 20 years ago. So shifting gears a little bit, you know, ablation is a very important modality, but all these patients get drugs, and the drug development has stalled. And you talk about AMIO, IV AMIOLOAD. We do have or will have very, very soon IV SOTOLOL, which is indicated for AF treatment. It doesn't have a VT indication as far as I know, but do you think that that will be, you know, another modality of treating VT storm for the patient? Do you think that will be dangerous? What do you think? Well, just a comment. So, you know, one of the other things we think of is, too, most of these patients have low ejection fractions and heart failure, and so some of the drugs, even when we're using them to treat AFib, you know, acutely it's nice to have the drug to, you know, to terminate AFib, but for VT I don't, I think, unfortunately we don't have a lot of drugs, but I don't see it having a major role in a large number of patients, at least the classic or the typical patient we treat with VT storm. I think that is a correct sentiment. There's one repurposing that is likely to happen in the field, and that is good old propranolol and nadolol are two drugs perhaps that are going to make a slight bit of a comeback, because they tend to actually work better in VT storm for multiple pharmacologic reasons, and the drug industry is now having some extra impetus to look at these sort of new types of potassium channels that are targeted and impacted by some of these drugs. So anytime we say a drug belongs to a certain class, it is not pure for that particular mechanism. It always has smaller effects, and we think propranolol is one drug which has been around for 40 years, 40 plus years, and that is actually likely to be repurposed and used a little bit more. But again, you're talking about this at the very end stage of disease. The bigger question is what can we do more upstream, and that's the one area where we think there's going to be a big impact. Two related areas. One is, of course, is inspired by cancer pain therapy groups, because they target the same signals that go from various organs to the brain, the pain pathways. The same signals going from the heart to the brain is why the heart develops scars. So we now know that if you infarct the heart and if you can cut those signals using new pharmacologic compounds, you can reduce fibrosis by 80%. So it's quite likely that you're going to have very powerful anti-heart failure treatments, which means if you don't have much scar, your VT burden comes down. So that is a big area of focus within the pharmaceutical industry, and we think that those types of treatments that are applied way early, we are talking about 10, 15 years before a patient comes in with structural heart VT, is likely to have an impact. Upstream therapy, yes. Any other novel drugs that anyone use? Anyone use rinolazine for heart VT? We have, and we're still not able to quantify its benefit. It certainly works for AFib. It's extremely sick, multi-ablation patients. But we are still not able to interpret our data with rinolazine. Have you used it? No, I mean, the big trial, there was a reduction in ventricular arrhythmia seen in the treated group, and it has an effect. Again, we have used it, particularly in combo with genetarone and amino, with AFib patients, and it seems to have maybe some benefit, but I have not used it in VT patients. Okay, talking about gap analysis, devices today are so sophisticated. And basically, you know, you can program a device almost any way you want. Is there any opportunity for doing better, or do we just have to be smarter about using the tools that are already available? I don't want to preempt your talk. I'll preempt my talk. How does that sound? So I help kind of organize our device clinic, and I think you probably all feel this pain, which is just this enormous deluge of data. And we, in some ways, asked for it, you know, rhythm analysis, information about what's happening with a patient 24-7, and it just has resulted in this enormous, I mean, what is large enough planet of data that we just can't possibly keep up with? I mean, even our center, which I would say is extremely, you know, sort of well-resourced, we can hire, you know, the people we, you know, need, and there just aren't enough people out there, you know. We are perpetually understaffed in our device clinic, and the ability to manage the data is, you know, not 100%. So, you know, help in whatever form in terms of being able to automate, you know, analysis of information that we get. If there could somehow, please, be like some United Nations of device companies that could come together and just agree on a uniform, you know, modality of data transmission, and even just warehousing the data, you know, in the same place. I can't tell you how much of our, I'm kind of ranting now, aren't I? So how much, I feel this pain every day. So I am literally pleading with you. I like that, the United Nations of device companies. So, you know, so much of the workday of this, like, incredibly intelligent, well-qualified device nurse, okay, who can, you know, look at this data and, you know, help us sort out what is the problem with the device, is spent clicking buttons, you know, and transferring the PDF from, you know, the programmer to Epic, then uploading it into the, you know, and it's just, it's such an enormous waste of time that just feels like it could be solved, you know, by you guys. So thank you for taking that on. So for those of you in the device industry, I know that the IS-1 lead standard took, like, 20 years for the competitors within the field to finally come together and say, okay, we'll all do it the same way. Is there any move afoot in the industry to come together for standardization, and where are we in that? Can someone inform us? You want to make a comment? Yeah, anyone who has more knowledge on this? This meeting is about you guys as much as us. Please. So in light of full conflict of interest, I'm Manish Ratham. I'm an EP in San Diego, and actually I'm a founder of a company called Geneva Health Solutions. So we actually were bought by BioTel last year, so now I'm the CMO of BioTel. So to answer your question on several levels, number one, the IDCO subcommittee has been working for about 12 years. David Slotweiner has headed that committee. That team has done a great job working with all the device manufacturers to create IDCO standards. We're actually now working with them as third-party vendors, the guys in my space, are now coordinating with the IDCO to now tell the IDCO what we need, because what we basically do is we're an intermediary. So, we're an intermediary between the device companies and the EMRs, irrespective of the EMR. So, all the companies working in our space are now working with the IDCO to say, this is the data we need, these are the things we need to do, and ultimately, I still practice medicine. We started this company eight years ago with the concept of fixing my own problem, so that's exactly what we do. So, this is exactly what we do in the DataDelU space, is we bring all the data, we name the company Geneva, because the device companies wouldn't play nicely in the sandbox, so we said we need a Switzerland, we need a neutral party, hence the name of the company. So, that's exactly what we do, is we bring that data down. And again, the idea is not to have Geneva, I'm just talking, there are options in this space, there are other vendors that compete in this space that work in this middle-level EMR space. Competition is good. Yeah. If you're alone, you're alone. If there are two or three, you actually have an industry, so that's exactly what academia should be doing. But there is also another important unmet need, if you were to step back and look at the field of EP, ultimately, any disease, it's not just our field experts sitting here on this podium. Human disease ultimately requires only a few variables. You need to know what is the incidence, prevalence. Do you understand what is really happening, so that you can do something about it, so that then you can look at outcomes. So, in our business, in our field, our incidence data is all over the place. And since we don't have a good handle on incidence and prevalence, our natural history data is also spotty. So, in every condition, every disease, you open the book, and when a patient sits with you, it's very hard to prognosticate. If you have something, what road is this patient going to walk on? FDA, at one time, and the NIH once felt that we'll one day have rivers of data in which we can actually float these little boats called clinical trials to understand what an intervention does. But, right now, it looks like Niagara Falls, and you're standing right underneath it. And you're not showing what, sure, what data is going by that's actually relevant to us. So, I think therein lies a great opportunity, both for academia and for industry. Right. And actually, I'll also make one other comment about this, is that we have to be one of the comment is the reason we, as intermediaries, are now working with the IDCO subcommittee on a higher level is because, for example, that data they'll use. Right now, we're using software services. We're using people. At the end of the day, a PUA of 2.4 is a PUA of 2.4. We can run AI on this. Right. So, actually, there's a company in France, Simplicity, that's one of the members of this group, that is doing a lot of work in that space, and then, obviously, every company is moving in that space on some level. So, I think this is going to get better over time, and that's the purpose of this interaction that we're now developing, is we're saying, listen, we need the data in this format so we can run AI, and now it's just a matter of getting all the device manufacturers to agree to give us that data, and then it just makes it a lot smoother for all of us. So in terms of needs analysis, data management is one. What else about devices? Size, battery life, functionality? Is there anything that we would like better than we have right now? Or have we truly transitioned to a commodity mode for high-energy devices? I think, before commenting on that, I think also you did comment, you can't find the people. So I think, you know, the training of personnel, we, I think, all of us at academic centers wind up training our own with nurses, techs, PAs, NPs. There's such a great need, and as there's more and more information coming at us from implanted devices, not even to mention the wearables, I don't know what everyone else is doing with that, but it's a huge amount of information, and the type of people need to be so specialized, and so the need is also a training, training programs and more training programs in different aspects of this arrhythmia management. And of course, the very best people that we train get hired by you guys the next week, so. So one comment, I want to sort of applaud the move towards sort of machine learning. I mean, kind of like when you read ECGs, and you read a ton of them, when the machine says it's normal, chances are it's normal, and you could focus your attention on those that say it's not normal. So we spend so much time looking at routine device downloads that a machine can do maybe better than the average person, and I think that ultimately, we need to be alerted about deviations from normal and not the, you know, I don't even know, 90% that really don't require high-level expertise. That point is well taken. What many people, it's not just our field, other fields are also struggling with the same information. Just because a machine classifies it a normal, what is a quality check, you know, on these algorithms? It's possible that there may be a point in time where if you actually have to go dig in to find out if there is something that went wrong, it may become very hard if you have layer upon layer of algorithms, and I think that's a challenge for our side of the game where we really probably should have interdisciplinary PhDs and MD-PhDs who can actually work on computer science plus medicine because we need to have some interlocutors in those worlds. You know, I don't think we learn much by being in our own world because other industries, in fact, a group that we greatly look up to are pilots. We learn more from pilots as to how to be a better person in the lab, in the procedures that we do. So there has to be some kind of cross-fertilization, and perhaps it's a conference like this that is very interesting for us because it's not the same old, you know, one group talking just in an echo chamber to itself. I think just commenting, I think what you're alluding to, for the devices themselves, and I don't think, I think we've obviously made dramatic steps forward and have great technology that does a lot of things and does, you know, it discriminates arrhythmias, you know, better and better. But we still, you know, we still have some issues, and so even, you know, more young patients are getting devices, still the majority of devices, you know, less than 5% are leadless devices for defibrillators that are being implanted in the U.S. But leads, you know, the lead technology, leads can fail. Younger patients are getting devices, their leads need to be extracted, and even the leadless systems, you know, or leadless pacemakers in particular, you know, what do you do with that in a younger person? So we still, we haven't completely solved it in, you know, extraction, so I think we still have a lot more in technology that we can do, but we've come just, I mean, in my career I just, it's amazing from epicardial devices to what we have today has been amazing. Right. So what about indications for primary prevention? We've been stalled on that threshold for, what, a couple decades, right? Yeah, yeah, and there have been a couple runs at refining that, you know, there's this population of patients, EF, you know, 37%, but high scar burden, you know they're at high risk, and just to inform the group, the standard approach of the standard clinician is to keep ordering ejection fraction measurements with different techniques until you get an answer that you like and then you use that one in your chart documentation to justify the implant, and I'm only half joking with that, but we know that there are a lot of patients at risk, we know that our risk stratification is not effective, there have been many attempts with various non-invasive risk stratifying modalities that have not panned out, where do you guys see the future going, or are we in this rut for the next 20 years as well? I mean, I guess sort of piggybacking off of what Dr. Shivkumar was saying earlier, I mean, if we had access to, you know, hundreds of thousands or even millions of patients in sort of this unified nationwide database, you know, there are certainly signals that deep learning could help us pick up, you know, to understand who are the people who actually benefit from these devices, who are the people with higher ejection fractions, you know, back to your slide, you know, who we aren't picking up and not saving, I mean, in terms of pie in the sky, I think that's, you know, that is where we can kind of solve this problem. So how do we get there? Yeah. I think it's two-pronged, David. My vote would be is you really have to look at it on two fronts. One is we really need a lot more basic science. So what amazed us is yesterday we had this research discussion with the Northwestern University team, Dr. Arora and Dr. Geist, some of their scientists, long recording data sets for weeks and months. We don't even have good mathematics and strategies to analyze those data sets. So we have all these little ice-pick point-in-time measures. So we currently, even if any of your companies over there can suddenly come and give us six months, every single heartbeat of a patient ECG, we're going to say, thumbs up, good luck. We have no way of even approaching those data sets. That's actually a big problem. So we need to actually solve that, and that is something that where we probably can have, you know, special RFAs that the NIH can put out, and that has to be done. And I would also encourage industry to really focus on bioengineering research partnership grants. The blessing of America, the blessing of the NIH is that we really is one of the very few countries in the entire world where public money is invested to create and expand industries. So many of you as scientists should be writing precisely for those types of grants to ask these types of questions. So collaborate with your academic partners or in your geographic area. You should go to Dr. Russo's lab and say, okay, we want to apply for this grant. And that is actually highly underutilized. So if you talk to some of the top executives in any of our scientific organizations like NSF, it's actually undersubscribed. And I was amazed to see that I've sat on a big panel for Canada. And in Canada, any of the public funds that are available, industry uses it very effectively. So there is a little bit of an opportunity here where there is a pot of money and there are sources that are available. So you don't have to just go to your marketing manager whose only view is the next 90 days and share price. So you can actually have a slightly longer view of the field. So I think it's going to require more partnership. So it's basic science and trying to make use of preexisting data. So those would be some kind of a position statement. We could partner with professional societies to get the word out. Great point. And the other thing, too, is we often, you know, years ago, we even started doing studies. What arrhythmias do you have that predict the bad arrhythmias, right? When you see all these PVCs and you see, you know, RNT, whatever you're seeing or that can predict. We see it all the time when we're, you know, starting someone on defetalide, on telemetry. If we start to see a lot of, you know, ectopy and, you know, some polymorphic beats, you know, we know we want to stop it, right? You know, they're in the hospital. We can predict that from ICDs. We can predict that actually even with AFib. These micro, we were talking this morning with someone here, with Christine, the micro AFib episodes can predict AFib. So if we can get better at that in predicting the arrhythmias before they happen, I think in using AI or using all these resources that we don't currently, you know, have available, I think that'll help us a lot in the long term. I think that the complexity of this question is so enormous that we have difficulties really knowing even in retrospect who died of an arrhythmia and who died of some other catastrophic event. And many of the things that predict arrhythmic death predict overall mortality as well. So even defining the endpoint is problematic. And those of us who have been around for anything more than a decade have, you know, survived all the enthusiasm and then disappointment about these non-invasive risk stratifiers. So T wave alternans is going to be it. Heart rate turbulence, mirror reflex sensitivity, signal average DCG, and no one does any of these, right? I think it's not going to be one answer. You know, it's going to be a combination of autonomics, imaging, lifestyle, genetics, and how we're going to accumulate that data is going to be challenging. And it's probably not going to be one risk assessment at one point of time because these things change over time. So it's a dynamic ongoing problem. And I don't think that this is going to be an industry-sponsored issue. This is a public health issue, and it's going to take an enormous investment from NIH and from sort of big data companies to get to the bottom of this. And I think, you know, with the consolidation of, you know, hospital systems and insurance companies, I mean, there are preexisting pools of, you know, millions of patients that, you know, might be a starting point for trying to collect this information. Do we have any questions out there in the audience? You guys have been awfully quiet. Yeah, can we have the microphone? In back, in back. Okay. I come from a tech industry and have no device and medical background. I've been getting exposed. You talk a lot about, in the past two days, precision and being more targeted. But when I look at a mapping system in a cat lab, it looks like an 18th century confusion going on. Is there any improvements to be made there or any of the process that can help you guys? In terms of mapping? I would just say I'm just using that as an example. Mapping may be one or anything else. I mean, I think the biggest change in the last several years is higher density mapping with catheters that have not just, you know, one electrode sort of touching various parts of the heart, but, you know, 64 electrode pairs mapping with higher density and with, you know, better sensitivity, you know, the substrate. I think that's been a game changer. I would say that I think there is an opportunity. We were discussing with someone in the industry yesterday. Again, going back to this machine learning, you know, so many of us pride ourselves, okay, when you see an SVT, what's the maneuver that's going to distinguish A from B? Or where is this PVC coming from? And we memorize these archaic algorithms that have, you know, modest sensitivity and specificities. These are machine learning issues, right? I mean, the PVC location, if we pulled our data from everyone who had an ablation for PVCs so that an independent EP in the middle of rural Arkansas could have the same accuracy and know exactly where to go as Dr. Shivkumar, Dr. Russo, or Dr. Kim, this is all open to us. We pride ourselves that we know these mechanisms, these algorithms, and we put them on boards, Dr. Haynes, who's going to be on the APIM. Truthfully, these are something that lends itself very easily to that sort of approach, but it takes an investment from academics and from industry. So I'm interested to see what you guys think of that. Is that an insult to our intelligence to say, hey, I want my pruka to tell me what I need to entrain next to distinguish ATAC from atypical AVNRT? What do you think? Yeah. No, I don't think it's an insult at all. I think you still need to understand and to make sure the data you're acquiring is accurate. So, you know, whoever's the machine or the person collecting, the cell person's, you know, standing there, you know, or sitting there collecting the data. So what we really want from you in tech industry is to say here's our map, and then just give us an arrow as to where to ablate or if it's a line, just to develop that based on all these things that you have from all these cases around the world. And I would even extend that one to a further point to say that some of the people actually practicing in Arkansas or a small town in Iowa sometimes actually do a lot of cases, and they've come up with very good practices. And I've been amazed when actually I go do cases in other labs where we pick up things by saying, oh, my God, this is so much better. We should incorporate that. So I think that kind of learning can also be improved if you actually have these kinds of systems that are put in place. Because after all, you know, they're our colleagues. They're people we trained at one point in time, and they've become very specialized in how they handle that. But they're, you know, sort of deluged with so much information. To make it easier for us would any day be a big plus. Another question in the back there? I'm going back to the discussion on devices and the technologies and training and so on, and the recognition that in some ways our devices have complexity to them that sort of we can miss how relatively simple they are, and that mostly they're doing what they need to do. And I actually, when I woke up this morning, I was thinking of the case from yesterday on the patient who finally it was a pretty damn simple fix, and an awful miss that it wasn't picked up much, much earlier. A fairly relatively simple thing related to PVAR for heaven's sakes. So I guess between ourselves and yourselves, it's we create too many layers. Even our own internal training, we can get sidetracked by some minor feature that doesn't really make a difference, yet we try to make it a sellable feature. So I guess it's not so much a question as a point about don't lose sight of the simple things and end up picking the wrong nail and using a very large hammer on it. And to that point, one of the things that personally makes me uncomfortable, and I realize it's here and it's even going to trend more as we use more machine learning to guide therapy, all of that stuff is black boxed. You really don't know the logic that the computer is using, the logic, to come to their decision, but they're using it on a data set of a million samples and the outcomes are relatively reliable. So we are trusting the machine more and more and getting used to the machine making decisions for us, and sometimes the kind of simple old school programming options get pushed aside and we let the machine decide. And that, as an old guy, that makes me uncomfortable, but that's what's happening. And I think there's a balance to be struck for sure. I mean, one issue is ambulatory cardiac monitoring. Now that we have the ability to collect and look at 24, 7, for 14 days or 30 days worth of heartbeats, I mean, there are an average of 100,000 heartbeats a day. I definitely don't want to look at all of them. And so one of the companies that we use has a deep learning algorithm, and you're right, we don't see everything, so we're putting some faith, there's a leap of faith, that they're analyzing things properly. But I think like Rod said, they're pretty good at calling normal, normal, and they'll show us the things that they think are abnormal. And it's not always correct, but, you know, at least there is this sort of check in place where we get to see the primary, you know, rhythm strip to decide whether we agree or not. And I think, you know, hopefully we don't abandon that piece of it, which is, you know, our own, I guess, primary review of at least some of the data. Great. Well, thank you, panel, and that was a good discussion.

ventricular tachycardia

sudden death prevention

patient safety

collaboration

data management

risk stratification

machine learning

technology integration