Friedman, Paul A.

Introduction: Ebstein’s anomaly (EA) is a rare congenital anomaly of the tricuspid valve associated with heart failure and arrhythmias. Currently available imaging modalities like transthoracic echocardiography, cardiac magnetic resonance imaging (MRI), or computerized tomography (CT) are not viable for screening large asymptomatic populations. Hypothesis: A deep learning algorithm, implemented using a convolutional neural network (CNN) applied to the standard 12-lead ECG can be used as an accurate screening tool for EA in children and adults. Method: We identified 1,154 EA patients and 21,430 age- and sex-matched controls (1:20 ratio) between 1987 - 2020. A CNN was trained, validated and tested using digital 12-lead ECGs in 70%, 10% and 20% of the cohort respectively, divided by random shuffling. Results: The mean age was 26.68±19.23 years (41.8% aged < 18 years, 56.41% female) for the EA cohort and 28.36±18.65 years (37.2% aged <18 years, 55.67% female) for the control cohort. Following training and validation the area under the curve (AUC) for detecting EA in the testing cohort was 0.976 (95% CI 0.962 - 0.989) with prediction accuracy of 0.959. In subgroup analysis, the AUC was comparable in males (0.970, 95% CI 0.946-0.995) and females (0.979, 95% CI 0.964-0.994), those <18 (0.980, 95% CI 0.958-1.000) and ≥18 years old (0.972, 95% CI 0.954-0.989), as well as in those with right bundle branch block (RBBB) (0.969, 95% CI 0.950-0.989) and without RBBB (0.950, 95% CI 0.922-0.978) on ECG. Conclusion: Artificial intelligence enhanced ECG based detection of EA is possible with high diagnostic accuracy. Following further refinement and external validation, this algorithm can facilitate early detection of EA using ECGs, a widely available and inexpensive tool. Figure: Receiver operating characteristic (ROC) curve for the performance of the artificial intelligence CNN in detecting EA in the test cohort stratified by (A) age (< 18 and ≥ 18 years) and sex and (B) presence of RBBB.

Introduction: Prior AF screening trials demonstrated low yield, highlighting the need for more targeted approaches. An AI algorithm was developed to identify ECG signatures of AF risk during normal sinus rhythm, which has been validated in diverse external populations. Hypothesis: An AI-guided, targeted screening approach could improve the diagnosis of AF. Methods We conducted a pragmatic decentralized trial to prospectively recruit patients with stroke risk factors but no prior AF. The AI algorithm was applied to the ECGs performed in routine practice and divided patients into high- or low-risk groups. The primary endpoint was AF lasting ≥ 30 seconds on a subsequent 30-day continuous cardiac rhythm monitor. In a secondary analysis, trial participants were 1:1 propensity score-matched to real-world controls derived from the eligible but unenrolled population. Results A total of 1,003 patients from 40 U.S. states completed the study, with a mean age of 74 [SD 8.8] years. Over a mean of 22.3 days of continuous monitoring, AF was detected in 6 (1.6%) of low-risk patients and 48 (7.6%) of high-risk patients (OR 4.98 [2.11-11.75], p<0.001). Compared to usual care, AI-guided AF screening was associated with increased detection of AF (high-risk group: 4.2% vs. 11.1%, p<0.001; low-risk group: 0.9% vs. 2.4%, p=0.12) over a median follow-up of 10 months. Conclusions A prospective pragmatic study found that the AI a can risk-stratify a relatively uniform population (i.e., older adults at risk for stroke) to detect AF during short-term cardiac monitoring. Furthermore, when compared with usual care, AI-guided cardiac monitoring was associated with increased AF detection. As such, an AI-guided AF screening approach, leveraging existing EHR data and infrastructure, could be effective, patient-centered, and massively scalable, thereby reducing unnecessary health utilization and diagnosis-related anxiety (Clinicaltrials.gov: NCT04208971).

Background: Recognition of the infectious characteristics among the sepsis-related early mortality after transvenous lead removal (TLR) is an essential step to improving patient outcomes. Objective: This study sought to understand the infectious characteristic and circumstances among the sepsis-related early mortality after TLR. Methods: We performed a retrospective analysis of all patients undergoing TLR from January 1, 2001, to January 1, 2021 at the Mayo Clinic (Rochester, Phoenix, and Jacksonville). Patient details were identified through a detailed chart review of the electronic health record from within the Mayo Clinic system and outside records when available. Results: A total of 2,319 patients were included in the study. The overall 30-day sepsis-related mortality rate was 1.25% (n=29). Among the sepsis-related deaths, 24 patients (83%) had systemic cardiac implantable electronic device (CIED) infection, and 4 patients (14%) initially had isolated pocket infection. The average time from admission to TLR was 11 days. MSSA was the most common organism (n=8, 28%), and this was followed by MRSA (n=4, 14%) and Candida (n=3, 10%). Twenty of these patients (69%) had vegetation on the echocardiogram: vegetation on lead only in 9 patients, vegetation on both valve and lead in 8 patients, and vegetation on valve only in 3 patients. Among valvular vegetations, 5 patients had tricuspid valve vegetations, 3 patients had aortic valve vegetations, 2 patients had mitral valve vegetations, and 1 patient had both mitral and aortic valve vegetation. The average size of vegetation was 1.3±0.8 x 1.1±0.6 cm. Conclusions: The 30-day sepsis-related mortality rate was 1.25%. Among the sepsis-related early deaths, MSSA was the most common organism, and more than two-thirds (69%) of the patients had vegetations on cardiac imaging. This highlights the importance of further research to improve the infection outcome in TLR patients.

Background: Recognition of the causes of early mortality after transvenous lead removal (TLR) is an essential step for the development of quality improvement programs. Objective: This study sought to determine the causes of early mortality after TLR and to further understand the circumstances surrounding death after TLR. Methods: We performed a retrospective analysis of all patients undergoing TLR from January 1, 2001, to January 1, 2021 at the Mayo Clinic (Rochester, Phoenix, and Jacksonville). Causes of death were identified through a detailed chart review of the electronic health record from within the Mayo Clinic system and outside records when available. The causes of death were further characterized based on whether it was related to the TLR procedure. Results: A total of 2,319 patients were included in the study. The overall 30-day all-cause mortality rate was 3%. Among all 30-day deaths, infection was the most common primary cause of death (42%). This was followed by decompensated heart failure (17%), procedure-related death (10%), sudden cardiac arrest (7%), and respiratory failure (6%). The 30-day mortality rate directly due to complications associated with the TLR procedure was 0.3% due to vascular tear and cardiac perforation. A third of deaths (33%) occurred after discharge from index hospitalization, among which 43% were readmitted before their death, 35% died at home or a nursing facility, and 22% were discharged on comfort care and died in hospice. The main reasons for readmission prior to death were sepsis and decompensated heart failure. Conclusions: The majority (90%) of 30-day mortality after TLR was not due to complications associated with TLR procedures. The primary causes were infection and decompensated heart failure. This highlights the importance of further research to reduce postoperative mortality from these factors, including deaths after hospital discharge.

AbstractIntroductionData regarding ventricular tachycardia (VT) or premature ventricular complex (PVC) ablation in patients with aortic valve (AV) intervention (AVI) is limited. Catheter ablation (CA) can be challenging given perivalvular substrate in the setting of prosthetic valves. We sought to investigate the characteristics, safety, and outcomes of CA in patients with prior AVI and ventricular arrhythmias (VA).MethodsWe identified consecutive patients with prior AVI (replacement or repair) who underwent CA for VT or PVC between 2013 and 2018. We investigated the mechanism of arrhythmia, ablation approach, perioperative complications, and outcomes.ResultsWe included 34 patients (88% men, mean age 64 ± 10.4 years, left ventricular (LV) ejection fraction 35.2 ± 15.0%) with prior AVI who underwent CA (22 VT; 12 PVC). LV access was obtained through trans‐septal approach in all patients except one patient who had percutaneous transapical access. One patient had combined retrograde aortic and trans‐septal approach. Scar‐related reentry was the dominant mechanism of induced VTs. Two patients had bundle branch reentry VTs. In the VT group, substrate mapping demonstrated heterogeneous scar that involved the peri‐AV area in 95%. Despite that, the site of successful ablation included the periaortic region only in 6 (27%) patients. In the PVC group, signal abnormalities consistent with scar in the periaortic area were noted in 4 (33%) patients. In 8 (67%) patients, the successful site of ablation was unrelated to the periaortic area. No procedure‐related complications occurred. The survival and recurrence‐free survival rate at 1 year tended to be lower in VT group than in PVC group (p = .06 and p = .05, respectively) with a 1‐year recurrence‐free survival rate of 52.8% and 91.7%, respectively. No arrhythmia‐related death was documented on long‐term follow‐up.ConclusionCA of VAs can be performed safely and effectively in patients with prior AVI.

Background. Mortality risk assessment before kidney transplantation (KT) is imperfect. An emerging risk factor for death in nontransplant populations is physiological age as determined by the application of artificial intelligence to the electrocardiogram (ECG). The aim of this study was to examine the relationship between ECG age and KT waitlist mortality. Methods. We applied a previously developed convolutional neural network to the ECGs of KT candidates evaluated 2014 to 2019 to determine ECG age. We used a Cox proportional hazard model to examine whether ECG age was associated with waitlist mortality. Results. Of the 2183 patients evaluated, 59.1% were male, 81.4% were white, and 11.4% died during follow-up. Mean ECG age was 59.0 ± 12.0 y and mean chronological age at ECG was 53.3 ± 13.6 y. After adjusting for chronological age, comorbidities, and other characteristics associated with mortality, each increase in ECG age of >10 y than the average ECG age for patients of a similar chronological age was associated with an increase in mortality risk (hazard ratio 3.59 per 10-y increase; 95% confidence interval, 2.06-5.72; P < 0.0001). Conclusions. ECG age is a risk factor for KT waitlist mortality. Determining ECG age through artificial intelligence may help guide risk-benefit assessment when evaluating candidates for KT.

Abstract Aims An artificial intelligence-enabled electrocardiogram (AI-ECG) is a promising tool to detect patients with aortic stenosis (AS) before developing symptoms. However, functional, structural, or haemodynamic components reflected in AI-ECG responsible for its detection are unknown. Methods and results The AI-ECG model that was developed at Mayo Clinic using a convolutional neural network to identify patients with moderate–severe AS was applied. In patients used as the testing group, the correlation between the AI-ECG probability of AS and echocardiographic parameters was investigated. This study included 102 926 patients (63.0 ± 16.3 years, 52% male), and 28 464 (27.7%) were identified as AS positive by AI-ECG. Older age, atrial fibrillation, hypertension, diabetes, coronary artery disease, and heart failure were more common in the positive AI-ECG group than in the negative group (P &lt; 0.001). The AI-ECG was correlated with aortic valve area (ρ = −0.48, R2 = 0.20), peak velocity (ρ = 0.22, R2 = 0.08), and mean pressure gradient (ρ = 0.35, R2 = 0.08). The AI-ECG also correlated with left ventricular (LV) mass index (ρ = 0.36, R2 = 0.13), E/e′ (ρ = 0.36, R2 = 0.12), and left atrium volume index (ρ = 0.42, R2 = 0.12). Neither LV ejection fraction nor stroke volume index had a significant correlation with the AI-ECG. Age correlated with the AI-ECG (ρ = 0.46, R2 = 0.22) and its correlation with echocardiography parameters was similar to that of the AI-ECG. Conclusion A combination of AS severity, diastolic dysfunction, and LV hypertrophy is reflected in the AI-ECG to detect AS. There seems to be a gradation of the cardiac anatomical/functional features in the model and its identification process of AS is multifactorial.

Abstract Aims Current non-invasive screening methods for cardiac allograft rejection have shown limited discrimination and are yet to be broadly integrated into heart transplant care. Given electrocardiogram (ECG) changes have been reported with severe cardiac allograft rejection, this study aimed to develop a deep-learning model, a form of artificial intelligence, to detect allograft rejection using the 12-lead ECG (AI-ECG). Methods and results Heart transplant recipients were identified across three Mayo Clinic sites between 1998 and 2021. Twelve-lead digital ECG data and endomyocardial biopsy results were extracted from medical records. Allograft rejection was defined as moderate or severe acute cellular rejection (ACR) based on International Society for Heart and Lung Transplantation guidelines. The extracted data (7590 unique ECG-biopsy pairs, belonging to 1427 patients) was partitioned into training (80%), validation (10%), and test sets (10%) such that each patient was included in only one partition. Model performance metrics were based on the test set (n = 140 patients; 758 ECG-biopsy pairs). The AI-ECG detected ACR with an area under the receiver operating curve (AUC) of 0.84 [95% confidence interval (CI): 0.78–0.90] and 95% (19/20; 95% CI: 75–100%) sensitivity. A prospective proof-of-concept screening study (n = 56; 97 ECG-biopsy pairs) showed the AI-ECG detected ACR with AUC = 0.78 (95% CI: 0.61–0.96) and 100% (2/2; 95% CI: 16–100%) sensitivity. Conclusion An AI-ECG model is effective for detection of moderate-to-severe ACR in heart transplant recipients. Our findings could improve transplant care by providing a rapid, non-invasive, and potentially remote screening option for cardiac allograft function.

AbstractIntroductionA current limitation of single chamber implantable cardioverter defibrillators (ICDs) is the lack of an atrial lead to reliably detect atrial fibrillation (AF) episodes. A novel ventricular based atrial fibrillation (VBAF) detection algorithm was created for single chamber ICDs to assess R‐R variability for detection of AF.MethodsPatients implanted with Visia AF™ ICDs were prospectively enrolled in the Medtronic Product Surveillance Registry from December 15, 2015 to January 23, 2019 and followed with at least 30 days of monitoring with the algorithm. Time to device‐detected daily burden of AF ≥ 6 min, ≥6 h, and ≥23 h were reported. Clinical actions after device‐detected AF were recorded.ResultsA total of 291 patients were enrolled with a mean follow‐up of 22.5 ± 7.9 months. Of these, 212 (73%) had no prior history of AF at device implant. However, 38% of these individuals had AF detected with the VBAF algorithm with daily burden of ≥6 min within two years of implant. In these 80 patients with newly detected AF by their ICD, 23 (29%) had a confirmed clinical diagnosis of AF by their provider. Of patients with a clinical diagnosis of AF, nine (39%) were newly placed on anticoagulation, including five of five (100%) patients having a burden >23 h.ConclusionsContinuous AF monitoring with the new VBAF algorithm permits early identification and actionable treatment for patients with undiagnosed AF that may improve patient outcomes.