All right, good morning to all of our distinguished guests, faculty, and friends. My name is Dr. Larry R. Jackson II, and on behalf of our wonderful collaborators, the AHA and Art Rhythm Society, I want to welcome everyone to Shark Tank 2025, Hooking Atrial Fibrillation. So our session format today, Dr. William Lewis will join the stage shortly. He will give a brief background on the Shark Tank program, talk about submission guidelines, reviewer criteria. We will have three separate 10-minute presentations followed by five minutes of Q&A where our sharks will attempt to draw blood. After the final Q&A section and when our sharks have done feasting, the judges will deliberate for about five to 10 minutes, we will select a winner, we will come back, present that winner, present gifts, certificates. So please stay, once again, please stay after the final presentation has been, and the gifts and awards have been distributed. So without further ado, Dr. William Lewis. I will repeat, there were many great submissions that we read. We chose three excellent presenters who will present their work today, so thank you. Good morning. My name is Bill Lewis. I'm the chief clinical strategy officer at MetroHealth Medical Center in Cleveland, Ohio. I had the opportunity to begin to work with the Get With The Guidelines program in 2013, initially launched by the American Heart Association. The program is a hospital-based quality improvement program. But the program is unique. It's unique because, first of all, it was the first Get With The Guidelines program to ever enroll outpatients, people in the emergency room or being treated in observation status. We developed a follow-up registry, which none of the other registries did have at the time. And in 2017, we did something rather unique. We actually collaborated with the Heart Rhythm Society. That kind of makes sense. But as you know, in societies, it's often very, very difficult to get organizations to collaborate on anything besides guidelines. We collaborate very well on guidelines, just never have had a program together. At the end of the day, despite the emergence of direct anticoagulants in 2010 and 11, adherence to anticoagulation in AFib was about 50 percent as late as 2016. The program has 250 hospitals participating and enrolling in about 48,000 patients annually. The program has been able to achieve a 99 percent adherence to anticoagulation, better than any other registry out there. So if you're not involved, you should get involved because this program is a way to demonstrate your improved quality of care. Fifty-eight hospitals were actually recognized by the Heart Association and Heart Rhythm Society last year for their quality performance. So again, being recognized, knowing what your capabilities are, is really important. The registry itself has generated 17 publications and has numerous papers in the pipeline. They discuss things in real-world scenarios that you can't discuss elsewhere, like it's difficult to do in randomized studies, such as off-label use of anticoagulants, anti-rhythmic drug patterns of care, things like that, that just don't get done in regular randomized trials. And it has been my pleasure to have been able to participate in the development and the operations of this program over the years. So without further ado, my next job is to introduce Dr. Javier Barcos, whose title of his talk is on the screen. First, atrial fibrillation ablation in patients age 75 years and older, findings of the Get with the Guidelines-A-Fib program. Dr. Barcos. Yes. Thank you so much. Good morning, everyone. I want to thank the HRS for this opportunity. It is truly an honor and a pleasure to be here. Today, I will present our proposal study entitled First Atrial Fibrillation Ablation in Patients Age 75 Years and Older, Findings from Get with the Guidelines-A-F is the most common sustained arrhythmia in adults. It is associated with an increased risk of death and multiple adverse outcomes. Moreover, it significantly impacts the quality of life and healthcare costs. The overall A-F burden has doubled over the past 30 years, with the USA among the top three affected countries. Three principal factors contribute to this trend, increasing prevalence of modifiable risk factors, improvements in surveillance and diagnosis, and progressive population aging. A-F prevalence rises with age. At the same time, global life expectancy has increased over several decades and now stands at 80 years in the USA. This Danish registry shows that A-F ablation have increased steadily over time. So catheter ablation has emerged as an effective treatment for symptomatic A-F, yet current guidelines class 1 recommendation for A-F ablation do not address elderly patients. However, these expert consensus statements provide class 2 indication for selected old and very old patients, aiming to improve symptoms and quality of life. Most randomized controlled trials enroll younger courts. Additionally, data on elderly patients mostly come from observational studies. These recent meta-analyses include more than 100,000 patients across 27 studies. However, only one was randomized. The meta-analyses demonstrate that long-term A-F recurrence was similar in both age groups. However, major complication rates were higher in the elderly, especially with radiofrequency energy. This is a sub-analysis of the Kavanaugh trial. Although catheter ablation was superior to drug therapy in reducing A-F recurrence at every age, its benefit on heart outcomes gradually declined and was lost in patients 75 years and older. The two main U.S. national registries about A-F ablation on elderly patients to date rely on ICD billing codes. So this probably has miscoding and lack granular clinical and ablation protocol details. Therefore, elderly patients are underrepresented in studies, underscoring the need to examine their characteristics, treatments, and outcomes. Ablation in elderly patients presents unique challenges. This population often exhibits multiple comorbidities and Freighter syndrome, requiring a careful risk-benefit balance. The objective of our proposed study is to compare clinical characteristics, treatment strategies, and ablation protocols, including procedural outcomes, among different age groups undergoing their first time A-F ablation. It also aims to deepen insight into the older group to guide more personalized treatment strategies. Our hypothesis is that patients aged 75 years and above undergoing their first A-F ablation have more cardiovascular risk factors, comorbidities, and Freighter than those aged 65 to 74 years, resulting in later stage ablation, longer procedures, different strategies, and higher complication rates. On this slide, you can see our proposed study design and methodology. It will be a retrospective multi-center analysis of the gateway guidelines of A-F registry. This program was developed to evaluate and improve the acute management of A-F in the USA, increasing adherence to guideline-based therapies and potentially improving outcomes. It also has usefulness for observational studies and the design of future interventions. This table summarizes the main features of this program. So first, we will conduct descriptive and comparative analysis of characteristics and procedural outcomes between groups. Then, we will use multivariable regression models to identify independent predictors of those outcomes. The inclusion criteria are patients aged 65 years and above undergoing their first A-F ablation with pulmonary vein isolation and whose primary diagnosis is A-F of any subtype. The exclusion criteria are any patient with previous A-F ablation or surgery, any ablation that doesn't include PBI, and any record with incomplete data or comfort-only care. After applying these criteria, we will include two age groups for analysis, the younger group aged 65 to 74 years and the older group aged 75 years and older. Our primary outcomes include clinical characteristics. As you can see, there are several items. But before moving on, I want to briefly clarify what is our approach to frailty. Although this database doesn't include a structured frailty assessment, we can use a frailty index and calculate using the deficit accumulation model. This database contains more than 30 variables that can be used, and there are validated cut-offs to classify patients. Another primary outcome is health-related social needs, such as education, employment, financial strain, and food access. The third item is echocardiographic parameters, for example, left atrial size and left ventricular function. The fourth item is pharmacological treatment, including antiarrhythmics and anticoagulants before admission and at discharge. The fifth item is ablation indication and protocol. As you can see, this database contains very detailed information. For example, periprocedural and interprocedural anticoagulation, anesthesia protocol, imaging and mapping use, transeptal approach, energy and catheter type use, including electroporation and ablation approach, among others. Finally, we will apply descriptive analysis and intergroup comparison to each primary outcome. Our secondary outcomes focus on periablation events that occur during the inpatient period. As you can see, the periablation variables are very detailed and can be grouped into efficacy and complication measures, as well as other procedural metrics. In this case, we will apply descriptive analysis, intergroup comparison, and multivariate modeling to these variables. This study has different or several strengths. For example, large nationwide registry, comprehensive clinical and procedural details, and real-world practice data. The limitations of this study are its observational and retrospective design, voluntary hospital participation, and follow-up limited to inpatient period. Thank you for your attention, and I am happy to answer any questions. Thank you, Dr. Barkos, for an excellent presentation. I will open the floor now to our sharks, and if you could just briefly introduce yourself before you ask any questions. Taya Glotzer from Hackensack, New Jersey. Great presentation, great idea. I had two questions. One, do you have the duration of AFib prior to the ablation? Is that a data point? I'm sorry, the duration? How long the patient has had AFib diagnosis? We propose different subtype of AF ablation, and then in this study, we can use different subgroup analysis. So you're relying on the clinician to say paroxysmal or persistent, but you don't know when the first, like have they had AFib for 10 years or one year? That's what I'm asking. I think the database doesn't include specifically that item, but for example, one variable that includes this database is the ECG RIN. Procedure. Yes, after the procedure, but also when the patient is ingressed to the hospital. Right. And then my final question is just, I hope you'll break it up older, 75 to 80, 80 to 85, and over 85, so we'll learn even more. Yes, there are several studies on different cut-off points of age, but I think we can do some subtypes analysis, but most of the evidence use this cut-off of age. Hi, Andrea Russo from Camden, New Jersey. Really nice presentation. I think it's a really important topic, important group of patients. We're doing ablation in older patients, and obviously with newer technologies, that may, I'm sure you're going to include the type of energy source, because obviously the impact on duration. The other thing I was wondering, you are just doing it as an acute, you're only proposing as an acute study. Did you consider matching to CMS data as an additional project? Because you have the age group of 65 or older, so you could potentially look at re-hospitalization or other outcomes. Yes, it's a good question. For my understanding, this database only includes inpatient patients or inpatient hospitalization, but I think that it could be possible to link with other database. Yeah, so that would be, and actually maybe the shark tank is just the inpatient side. Is that true? I don't know if it only, but you can link through. Yes, with other database. So, remission, yes, it is possible to do another study of follow-up complication and efficacy. And outcomes. Yes. I think that's the point. Efficacy and complication outcomes. Hi, Brad Knight at Northwestern in Chicago. I want to spend a little bit of time talking about the one thing you did, which was to include only first-time procedures. So, a lot of the data on age-related complications showing that it's safe or not safe in older patients comes from sub studies of large clinical trials that are usually industry sponsored device trials they're all always first-time procedures and that's kind of how you design this trial but here's an opportunity to look at redos so it would be interesting to know if most of these patients over 75 or 80 are having redo procedures unlike the younger people it'll be powered to look at complication rates it'll be more I think in line with the theme of the database which is real-life applications and none of the redo cases are included in any of these device trials so I would encourage you to either have a sub study or maybe open that up to all procedures not just even PVI but people who come in for redo just posterior wall isolation all kinds of new approaches for redos third redos that's the power of this database it's very interesting to two points to to comment one I don't know exactly if this database could link the readmission of the same patient I don't know if this database will analyze the same patient in the follow-up so in our methodology only include the first time for that reason and the second point is in some studies they observe that when an elderly patient has a recurrence of AF the the treatment in client it's more to a non-invasive second ablation so in some studies they found that the younger patients are more easy to do a second ablation and yeah I'm not sure I can address the first point you made but the second point is that I think this is we want to know what is actually going on out there so if that's the case that would be interesting to know that yes it's an open window is it very interesting yes so I know we know we have probably have to move on but I first of all the database includes first-time AFib so that's the only real indicator whether we have how long the AFib has been going on so that's the only way we know that the that's the only piece of information we have that will actually address that so let me ask this question do you think that patients who are over the age of 75 who undergo ablation are healthier than those people who do not from the same age yeah so people who are over 75 who undergo ablation are they healthier than people who are over the age of 75 who do not it's probably a selected a court so there's some referral bias there how do you think you can minimize the impact of that referral bias to balance the comparison between age groups there are several statistical approach like multiple regression model and propensity score but the population the elderly population that doesn't go to procedure it will it will doesn't include in this study but you have a database in front of you the answer the question right it will be compared at the difference between the group that undergoes ablation versus the characteristics of the people who do not undergo ablation and I think that's insightful because you may see differences in demographics comorbidities to dr. Lewis's point yes it is probably that that population were more or have more comorbidities I'm afraid but it is interesting that this is static will facilitate the prognostic factors inside the older group to establish what patient will be have benefits of the ablation I think that is important okay Megan tertia from Columbia thank you for a great presentation it follows up on dr. Lewis's question I was happy to see that you're addressing frailty and thinking about that because that seems to be a really important factor maybe even more important than your actual chronological age so calculating frailty indexes can be complicated because it relies on a lot of data and it looks like you were using the Rockwood frailty index which requires I think like 30 out of 36 variables to be complete for it to be valid I was just wondering if you could address what you might do if there's a lot of missing data other ways of capturing that construct of frailty well may seem that data to calculate frailty index well to amazing missing data and it can apply different statistical tools to for example and something it is important that I think that the this registry platform contains some checks and a mandatory fields that the different center have to complete and I want to additionally say that it's very important the age and I was very surprised of the importance of the frailty index because there are some study I didn't have the time to show but there are some studies that show that age is not so important that frailty the frailty population are is more probably that has a recurrence or a complication yeah thank you dr. Barkos thank you excellent job thank you our next presenter will be dr. Harry Shritharan who will be presenting precision AF profiling real-world energy choices to improve success and optimizing nationwide atrial fibrillation ablation procedures yeah hi my name is Harry thank you for the opportunity to present today I'm excited to introduce to you precision AF an innovative data-driven approach to refine our energy selection for atrial fibrillation ablation procedures AF affects millions of people in America and worldwide and its prevalence is only rising we know that AF is a heterogeneous disease process and that the underlying atrial substrate varies from person to person based on the individualized risk factor profile catheter ablation has emerged as one of our most effective strategies in managing AF and in this contemporary era we have multiple energy sources at our disposal namely cryo ablation radiofrequency ablation and the newest kid on the block pulse field ablation but the selection of which one we use remains largely driven by operator preference and training rather than objective evidence if we had the same patient go to different operators at different institutions they may have their AF ablation done with a different energy source this subjective approach may affect procedural success rates and patient outcomes and represents a significant gap in our clinical approach this is somewhat driven by a gap in our knowledge let's take two patients an example we have a 55 year old obese male with paroxysmal AF and hypertension and then we have a 72 year old diabetic female with persistent AF and a moderately dilated left atrium cryo RF or PFA which one would you choose does a one-size-fits-all approach work best or would a more refined approach be required in EP we pride ourselves on a precise and data-driven approach yet our selection of the fundamental tool for our procedures the energy source remains largely subjective position AF aims to challenge that paradigm and to refine our operator dependent approach to a more patient centered one position AF has three aims firstly to perform a comprehensive comparison of outcomes across the different energy sources secondly to identify the specific patient characteristics associated with optimal outcomes for each energy source and third and finally to develop a user friendly predictive model for energy source selection to aid real-time clinical decision-making this study includes all adult patients undergoing catheter based AF ablation in the get with the guidelines AFib registry it excludes patients who have undergone surgical ablation hybrid procedures and patients with missing energy and or catheter type the primary outcome is first-pass PBI success which is defined as both entrance and exit block the secondary outcomes are procedural duration metrics acute complications both as a composite and an individualized baked breakdown composite and the need for additional ablation strategies to further secondary outcomes if within patient data linkages available either within the registry itself or externally I repeat ablation procedures and AF hospitalization to address our first aim we've developed a machine learning based propensity score framework firstly we'll classify all included AF ablation patients into three energy modality groups of RF cryo and PFA with respect to sample size estimation to detect a 7% difference in success rate with 80% power we need 144 patients per group after adjusting for propensity score methods and missing data that's expected this increases to 443 patients per group across equal groups of RF cryo and PFA given our expected distribution of energy sources of approximately in the registry about 60% RF 30% cryo and 10% PFA we will need a total of 4430 patients to run the study there's a publication in 2021 from this registry with 5356 patients across approximately four years so we'll have more than enough patients to achieve this especially if that trend is expected to continue with 13,500 patients overall after this grouping we'll employ a propensity score estimation approach using gradient booster decision trees and this is a machine learning framework that captures complex nonlinear relationships unlike traditional regression before adjustment we would expect there to be substantial imbalance in our baseline characteristics and this should diminish after applying our machine learning derived weights ensuring that we're comparing similar patients across all three energy sources our final propensity score analytical framework involves three complementary approaches the primary analysis is inverse probability treatment weighting and the secondary analyses include firstly pairwise matching and then machine learning based doubly robust estimation which is specifically there to protect against model misspecification to address our second and third aims we'll analyze all patients undergoing ablation with RF cryo or PFA in those three distinct groups that's RF cryo and PFA distinctly we'll divide these patients in those three groups into a training set and test set with an 80% and 20% split respectively we'll then implement eight supervised AI or machine learning based algorithms to develop different models for each of the outcomes that I mentioned earlier we'll internally validate the performance of these models on the test set looking at two key metrics area of the curve score with highest most desirable and number of variables with the lowest most desirable from a practicality perspective the best performing models based on these two metrics will be chosen as the final model for precision AF to facilitate clinical translation we'll develop a user-friendly web application that can be used in real time with respect to feasibility position AF is highly feasible and can be completed within the 9 to 12 month time frame is only get with the guidelines a for registry data and requires no additional data linkage outside the registry itself unless it's required for those further secondary outcomes the study goals also align with the HR SNA is a submission for supporting quality improvement to specifically to expand further on feasibility with respect to my own experience and skills that my team's experience and skill set my PhD focus on development and clinical translation of prediction tools for cardiovascular disease with machine learning specifically on the right an example of an online tool that developed one of my projects and you can see variables input onto the left and the output as percentages or an individualized survival curves on the right too I'm an early career B fellow multiple first publications in this field of advanced predictive analytics in cardiovascular disease and my group has experienced running larger prospective trials and registry brace trials also we believe I experience a skill set gives us a good foundation to run this in summary position AF aims to compare outcomes across different AF ablation energy sources using a machine learning based propensity score framework it also aims to develop a user-friendly prediction tool to aid clinical decision-making and energy source selection it's highly feasible and achievable novel and relevant given the multiple energy sources that we have available right now in this contemporary especially thank you for your time fantastic presentation and expertly delivered questions from our sharks this is probably one of the most important questions right now in EP I was excited to see this so I congratulate you for looking at this energy source issue a few comments though a limitation of this database as I understand is that it's PFA cryo RF there's no PFA catheter design and so as you know RF ablation has become relatively standardized across different companies the cryo balloon is largely one manufacturer and so it's pretty easy to compare I'm not sure how you'll handle the fact that there's now for FDA approved PFA systems and by the time the start study gets started there will be multiple catheters available and so I want to know how you think about that the second thing is it may not be as predictors of which energy source it may not be the physician's choice for the patient's choice it's largely driven by access to this technology in the hospital and now is the time to do a study like this because it might be useful to find some important differences on the other hand it might be that in a year or two everybody's doing PF because that's what's available to address your first point which is completely valid this registry and database only has PFA as it's as a modality alone not the catheter type and that'll be a true limitation of this study and based on the data available I think there's no way to overcome that limitation to address your second question about availability of the energy modalities at different centers that's the reason why I've sort of developed the second third aims framework in the way that they are where there's actually individual the groups are studied distinctly as RF cryo PFA so if you only had cry RF or PFA available at your institution you could just look at the risk the success rates and the outcomes valid only to that I think most institutions will have to at least two energy sources available to go to your point about adoption of PFA sort of being the dominant modality going forward my center just got PFA a month ago and we're liking you a lot I suspect it will probably head that way too but in the end with anything that you adopt early, there will always be cases where you can't use it as well. And we'll find out more about that, I think, as time goes on. And that's where I think something like this could help from a decision tool perspective. So sure, it's a good question to have now, but I think it could be helpful later on also. And could support further research interests going forward in that area, too. Really nice presentation. In addition, my thought was similar, is that availability of whatever equipment is obviously what you choose to do. The other thing is I'm wondering, at least I'm thinking, using PFA, you have, I think, a lower threshold, maybe, to do more than you would normally with RF and not worrying about the esophagus. And how will you take that into account? Just because the technology's there, you see some scarring, you think, well, it's kind of recently persistent. Maybe I will do more. How will you account for that? So one of the outcomes is the use of additional ablation strategies outside of the pulmonary veins alone. And that's how we'll account for that. We'll look at that specifically. I think it would be an interesting question to ask, too. Are we doing more with PFA than we are with cryo-RF? Because we have the ability to do that. And I think the question that we haven't answered in that space specifically is, does that help more? Is it that we're getting more transmural lesions, perhaps, than we are with RF? And that may be the reason why posterior wall isolation, in addition to PBI, hasn't shown as much of a signal so far especially in that capillary RCT, which is Australian, too. I think mine is more of a comment. I applaud the use of AI. I don't think we've had a proposal come across the queue that has done that. I always worry with AI, sort of this generalizability and what goes in your algorithms. And there are some limitations with the guidelines with respect to diversity, race, ethnicity, and geography. So more of a comment, just to think that that may be a limitation. But I do really applaud the effort. That will definitely be a limitation for this. 90% of the patients included in this 2021 study were white. And then there was a substantially lesser proportion when you fractionate them amongst the other minorities, too. The propensity matching aims to address that for at least the first aim there, without being perfect at doing it. But it will sort of help with that, at least from a statistical standpoint. And then from a machine learning perspective, that's something that we always encounter, which is, how do you generalize this to minority populations more than anything else that aren't represented in registry datas? And the way that we do that is you can do certain statistical methodologies to do that, too, including one called adversarial de-biasing, where you try and penalize. Oh, you can use a framework called adversarial de-biasing, where you try and penalize the model if you can correctly identify those minorities that are through it. So there are ways to address that. But again, you can't overcome the fundamental part, which is that the registry is heavily skewed towards 90% of people were white, and the 70% of people were male, too. I had a follow-up comment on the machine learning aspect. So, and it's more of a comment as well. I think thinking about how you're initially going to be developing a predictive model that will be based on the entirety of the data available in the registry. But as we think towards the clinical implementation, which I think is really exciting, thinking about the next step of work that's going to be needed for additional model validation at the local level, where you're gonna have additional constraints based on data available for that site, availability of different technology and approaches at different sites, as my colleagues have mentioned. So just wanted to encourage you to think also about the additional work needed for model validation and de-biasing when you move towards clinical implementation. Yeah, definitely an important part of model development there. And I think that's why I wanna try and create these models, as less variables as you can, because you have to think about the poor people putting their input data into it too. And so, that's how I'm gonna do it. Okay, he's gonna do it, he's gonna do the talk, okay. Dr. Srinath on. Excellent presentation, and thank you. Thanks. Thank you. Our last and final presenter will be Dr. Dacoon Sun, who will present temporal trends in the use of rhythm versus rate control in patients hospitalized with atrial fibrillation and heart failure. Good morning, thank you for the opportunity to present our research idea in Shark Tank this year. My name is Dao Kun, I'm a third year PhD student from the University of Minnesota and today I would like to share with you our research idea entitled Variation in the Use of Rhythm Control in Patients Hospitalized with AFib and Heart Failure. As many of us know, significant evolution has occurred over the past decade in the clinical guidelines recommended treatment strategy for AFib in patients with heart failure. Rhythm control strategies are more strongly endorsed now in the 2023 guidelines compared to those from the 2014. In fact, catheter ablation for appropriately selected patients with HFRAF has been elevated from a class 2B to a class 1 recommendation. But there are still many questions to be answered regarding the nationwide clinical practice pattern in the use of rhythm control. For example, at the patient level, which factors most strongly influence physician's decision to pursue rhythm control in patients with AF and heart failure? We know all those important clinical factors, but what about social demographic factors? Does insurance status influence physician's decision? And how important are those social demographic factors compared to those clinical factors? And the second question is, at the hospital level, how substantially does the use of rhythm control vary across different hospitals in patients with AF and heart failure? The third question is, how has the use of rhythm control evolved over time for patients with AF and heart failure? So to address these questions, we propose to conduct a new study with three aims. To address the first question, we will use machine learning techniques to identify key predictors influencing clinical decision making on rhythm control in patients with AF and heart failure. And to address the second question regarding the use of rhythm control variation across different hospitals, we will describe the hospital level variation in the use of rhythm control in this patient population. To address the third question, we will evaluate the temporal trend in the use of rhythm control in patients with AFib and heart failure. So to define our study population, we will include patients with a primary or secondary diagnosis of AFib and a history of heart failure between 2013 and 2023. We will exclude patients without AFib, patients without heart failure. And we also will exclude hospitals with less than 30 admissions in the study population. Our outcome of interest is rhythm control, which is defined as the occurrence of one of the followings. Patients with in-hospital antiarrhythmic drug therapy, patients with in-hospital cardioversion, or patients with in-hospital catheter ablation for AFib. Or those patients with antiarrhythmic drug prescribed at hospital discharge. Or patients with future rhythm control strategies planned at the time of discharge will be defined as having the outcome. To address the first aim, we will use machine learning approach and we will consider the followings as potential predictors. These are the medical history and clinical variables, the demographic characteristics, and very importantly, the social economic factors. We will use random forest to first explore any potential nonlinear association. And for the primary analysis, we will use logistic regression with least absolute shrinkage and selection operator known as the Alesso approach. For the second aim, we will describe the proportion of patients receiving rhythm control at each hospital and describe the pattern nationwide. Next, we will use adjusted approach by fit a multi-level logistic regression model with a random intercept. This model will account for the confounding fact from patient factors that we just identified for aim one. And this model will also account for the clustering within hospitals. We will calculate median odds ratio. This parameter will help to answer a very important question. That is, if the same patient with AFib and heart failure were admitted to two randomly selected hospitals, the median increase in the odds of rhythm control at the hospital with higher propensity to use rhythm control is how many times higher compared to the hospital with lower propensity. For aim three, we will first describe the proportion of patients receiving rhythm control across calendar time. Next, we will use adjusted approach by fit a multi-level logistic regression with a random intercept for hospital. And this time, our predictor is the calendar time. And we want to know after adjusting for patient factors, after accounting for the clustering within hospitals, is calendar time still significantly associated with outcome? And how does it look? So in summary, we hope our study will help to identify patient, provider, and system level factors that drive the choice of rhythm control in patients with AFib and heart failure. We want to quantify the differences in rhythm control adoption across hospitals, potentially revealing disparities in care access, institutional protocols, or resource availability. We hope to provide evidence to assess whether clinical practice aligns with involving evidence. And we hope our results can help us better understand the nationwide clinical practice pattern in the use of rhythm control. And that's all I got for today. Thank you all for your attention. I have a quick question. So very nice presentation, first of all, and that's important, I mean, with all the data we have in rhythm control in this cohort. Were you planning on looking at, you know, it might vary based on the type of heart failure, heart failure reduced EF or preserved or, you know, mid EF. Would you be looking at that differently or, you know, as part of one of the variables? Right. So that's a very important question, because what I just described are the primary analysis. Depends on the sample size, we would very like to do sensitivity analysis within subgroups. Because now we know the evidence showing the usefulness of rhythm control probably vary a little bit between subgroups, for example, half-REF versus half-REF. The evidence are not equal. So if we have enough sample size and statistical power to explore the question, we would very much want to do that in the subgroup analysis. I don't really have a question, I just want to congratulate you on looking at a topic that's very important. So I'm going to read a text I got this morning. Brad, my mom just got admitted with recurrent AFib and heart failure to your hospital. Please give me a call. Like this is a really big problem. So, you know, I think the half-REF, half-REF issue is going to be important. But I think you impressively showed how much thought you put into the data analysis and the statistical approach that you're going to take, so, thanks. Thanks. Very nice presentation. You mentioned that the 2023 guidelines really directly, more directly address heart failure, comorbidity. I was wondering if you think that might impact practice and whether that means there might not be enough time yet to look at temporal trends if it's too soon. Right. So the most recent evidence is from the 2023, while in our data study population, we want to include patients by 2023. But the question is, by the time or before the clinical guidelines are published, are our physicians already starting to adopt the rhythm control due to the existence of evolving evidence? So there will be a time gap, but we believe potentially many physicians already started to adopt the most useful approach for our patients, even before the publication of the most recent guidelines. Thank you. So you're really looking at rhythm control versus no rhythm control. Right. But I guess the other thing to look at is the rhythm control strategy that was chosen for the different groups. And if one rhythm control strategy failed, another rhythm control strategy, but you only have the acute data, right? Right. You don't have follow-up. Right. But again, I would imagine different subgroups, like older people might have one rhythm control strategy and younger people might have a different one. So you'll look at that. Right. So obviously the presentation's changed a lot from the abstract. So a couple of questions. So first of all, are you still excluding people with first-time AF and first-time heart failure? So that was my initial thought. But then I read a most recent publication from our registry, and it turns out that if my estimation was roughly correct, we hope to have about 20,000 to 30,000 patients with AF and heart failure. So the question is, if we only focus on patients without first AF or with first AF, the sample size will obviously shrink. It will be smaller. But then what I was thinking about is how to make good use of this data. So there are two options. First of all, we can do subgroup analysis only focusing on patients with a specific condition for them with first AF or without first AF. The second option is we can adjust for the status of the AF or the timing of AF as a covariate or as a confounder. And that could also be put in the analysis for our aim one, which is to identify the key predictors for our outcome. So these are the two potential options that I thought we could use. Yeah, because I would include them, because I think that it's not – I would include them. And the last thought is that the emphasis that you had on half-ref versus half-pef, and I know you addressed this already, but I would encourage you to really continue to look at that difference, because I think that what we're finding is that ablation can be very beneficial in patients with preserved ejection fractions and heart failure. Thank you. Thank you, Dow Kuhn. Great job.

Shark Tank 2025

Atrial Fibrillation

Dr. Larry R. Jackson II

Get With The Guidelines

AF ablation

Precision AF

machine learning

rhythm control

heart failure