Atrioventricular nodal reentrant tachycardia (AVNRT) is the most prevalent type of supraventricular tachycardia (SVT) in pediatric arrhythmia, resulting from dual AV nodal physiology substrate. However, accurately diagnosing it is challenging due to the absence of distinctive markers on baseline electrocardiograms. This study investigates advanced and hybrid ensemble machine learning (ML) techniques for detecting AVNRT in pediatric populations. A retrospective dataset of 263 subjects with 173 AVNRT and 93 controls was assembled. ECG data was processed to extract 12*65 quantitative biomarkers. Data preprocessing included dropping incomplete records, standardization, and label encoding. The dataset was split into 70% train, 15% validation, and 15% test, maintaining class uniformity. Synthetic Minority Over-sampling Technique was applied during training to address class imbalance. A wide range of 9 ML models (like SVM, KNN, XGBoost, 1D-convolutional neural networks (CNN), etc.) were evaluated on individual leads. The 2D models included baseline 2D-CNN, 2D-CNN+attention using Squeeze-and-Excitation blocks, and transformer architectures. Finally, hybrid ensemble models used majority voting amongst top performers during train-validation splits. The results indicate that leads aV RL, V 1, and V 4 are most informative, with individual models achieving an area under the receiver operating characteristic curve (AUC) of up to 0.85, while ensemble approaches significantly improved performance to AUC of 0.97. Furthermore, 2D-CNNs with attention-mechanism boosted performance, while transformers underperformed. The study highlights the promise of hybrid ensemble ML approaches as non-invasive diagnostic tools for the early detection of SVT in pediatric patients. Further external and prospective validation is needed to assess clinical acceptability to improve patient outcomes.

Background: Atrioventricular nodal reentrant tachycardia (AVNRT) is a common form of supraventricular tachycardia (SVT) in childhood resulting from dual AV nodal physiology (DAVNP) substrate. Children with AVNRT present with episodic chest discomfort or palpitations. The baseline electrocardiogram (ECG) in children with AVNRT appears normal. The ability to identify AVNRT risk from a baseline ECG may help during evaluation of children for palpitations with otherwise normal ECG. Hypothesis: We hypothesized that machine learning (ML) algorithms would be able to identify subtle ECG markers of DAVNP substrate on baseline ECG. Methods: Retrospective single center case-control study from 2020 to 2025, including 12 lead ECG data from children 5-18 years of age who had a confirmed diagnosis of AVNRT through electrophysiological study. Patients with congenital heart disease were excluded. The control group consisted of children 5-18 years presenting to clinic for chest pain or palpitations and had normal cardiac findings. ECGs were processed to extract 12×65 biomarkers. Preprocessing included normalization, handling categorical and missing values. SMOTE was used during training to prevent bias. The dataset was split into 70% train, 15% validation, and 15% test. Both 1D and 2D lead data were assessed during model building. The models included 1 dimensional and 2 dimensional convolutional neural networks (CNN), Support Vector Machines (SVM), K-Nearest Neighbors (KNN), Logistic Regression, and Boosting techniques. We implemented 10 ML models for 1D data: 9 traditional and a 1D CNN and used 2D CNNs for 2D data. Three ensemble models were also built using majority voting from the top performers. Optimal hyperparameters were determined using Bayesian optimization integrated with a grid search approach on the train/validation and evaluated on the test set. We used AUC, accuracy, sensitivity, and specificity to evaluate performance and compared the best models for statistical significance. Results: The case group included 170 children, control group included 93. Models using individual leads aVRL, V1, and V4 showed promising results with AUC as high as 0.85. Consistent improvement is observed by using ensemble models combining three leads in both traditional ML and CNN approaches, with AUC reaching 0.95 and maintaining balanced sensitivity and specificity of 0.85 each. Conclusion: ML algorithms can successfully identify subtle findings of DAVNP on baseline ECG that predict AVNRT risk.

BACKGROUND:The optimal pacemaker programming strategy for infants with isolated congenital complete atrioventricular block (CCAVB) remains unresolved. Dual-chamber pacing maintains atrioventricular synchrony and physiological heart rate variability but increases the burden of ventricular pacing on a myocardium that may be inherently prone to left ventricular (LV) dysfunction. OBJECTIVES:This study sought to compare clinical outcomes of dual (DDD)- vs single (VVI)- chamber pacing in infants with CCAVB (DAVINCHI). METHODS:A multicenter retrospective study (2006-2023) identified infants with CCAVB and pacemaker implant at <1 year, with single-site ventricular pacing and no significant congenital heart disease. Outcome measured were clinically significant LV dysfunction, mortality, and complications. RESULTS:A total of 109 infants (64% autoimmune CCAVB) were identified, 60.6% had VVI pacing. Over a median follow-up of 5 years, 60 complications occurred in 47 subjects (43.1%). Smaller infants had more complications. Clinically significant LV dysfunction developed in 11 (10.1%) and was more frequent in DDD (21% vs 3%; P = 0.006). LV dysfunction resulted in mortality in 1 patient and 10 patients required a change in pacing mode. Independent risk factors for LV dysfunction were DDD pacing and neonatal implant. Right ventricular pacing lead placement had a higher HR (HR: 2.67) for LV dysfunction but was not statistically significant (P = 0.20). CONCLUSIONS:DDD pacing increases LV dysfunction risk compared with VVI in infants with CCAVB. Single-chamber LV apical pacing should be considered in infants with isolated CCAVB who require pacing. There is a high risk of pacing-related complications, particularly with an increased risk of ventricular lead complications in low-weight neonates.

Background: Ivabradine is a hyperpolarization-activated cyclic nucleotide-gated channel blocker that has an off-label indication for arrhythmias in pediatric patients. Objectives: The primary objective was identification of the line of therapy and type of arrhythmia for which ivabradine was initiated. The secondary objective was to evaluate the initial and peak ivabradine doses, tolerance in partially fed patients (i.e., patients not receiving full feeds at the time of ivabradine initiation), number of patients continued home on ivabradine, concomitant drug interactions with a class C, D, or X classification, and adverse effects. Methods: A retrospective study of hospitalized patients <18 years of age receiving ivabradine for arrhythmias between January 1, 2021 and November 30, 2023. Data variables included demographics, ivabradine dosing regimens, concomitant anti-arrhythmic medications, percentage of ivabradine course while receiving full enteral feeds, concomitant medications with drug-drug interactions, and adverse effects (bradycardia [defined as <100 beats/min in neonates and <50 beats/min in infants and children], atrial fibrillation, vision changes, and emesis attributed to ivabradine). Results: Seventeen patients were included. The most frequent arrhythmia indication was atrial tachycardia and ivabradine was most frequently used as a third-line anti-arrhythmic. Ten (58.8%) received ivabradine despite not being on full enteral feeds. The median initial and peak ivabradine doses were 0.05 mg/kg/dose and 0.07 mg/kg/dose, respectively, and it was most frequently administered every 12 hours. Only 1 patient required a dose decrease for a noted adverse effect. Most patients (94.1%) had a Class C drug-drug interaction, and 3 had a Class X interaction. Ten (58.8%) were discharged home on ivabradine. Conclusions and relevance: Ivabradine was safe and effective as additive therapy for the treatment of pediatric arrhythmias. Larger studies of ivabradine use in pediatric arrhythmias are needed.

BACKGROUND:Atriofascicular fibers (AFFs) are rare accessory pathways that have higher rates of recurrence after ablation because of either failure to identify AFF (M) potentials or mechanical termination with contact. OBJECTIVE:We aimed to evaluate whether electroanatomic mapping (EAM) using multielectrode, high-density nonlinear catheters can reliably localize AFF potentials and determine a site for ablation without causing mechanical termination. METHODS:Seven patients underwent electrophysiology studies (EPS) and EAM using high-density, multielectrode catheters for antidromic tachycardia using AFFs. RESULTS:Mean age at ablation was 15.5 ± 7.4 years, and 2 (29%) were female. Two had Ebstein's anomaly. Three of 7 (43%) had previous ablation attempts. Of the previous attempts, 3 (60%) were complicated by mechanical termination during mapping. None of the previous ablation attempts was performed using high-density mapping catheters. HD Grid with Ensite NavX (Abbott Cardiovascular, Abbott Park, IL) was used in 3; the remainder used the PentaRay (3) or Optrell (1) catheters with CARTO (Biosense Webster, Irvine, CA). The AFF course was mapped to the lateral tricuspid annulus in all. Four of 7 had oblique courses with atrial insertion superior to the ventricular insertion. Multielectrode mapping catheters did not result in mechanical termination in any of these cases, whereas mechanical termination with ablation catheters occurred in 4 (57%). Successful ablation was performed in all with no recurrence over a 15 ± 6-month follow-up. CONCLUSION:EAM using high-density multielectrode mapping catheters allow localization of atriofascicular pathways without causing mechanical termination and may improve long-term success of ablation. Majority of the pathways had oblique courses with atrial insertion superior to the ventricular insertion.

Background Selection of radiofrequency ablation (RF) or cryoablation (Cryo) for atrioventricular nodal re-entrant tachycardia (AVNRT) in children remains controversial due to a lack of contemporary comparison studies in this population. Objectives This study sought to compare outcomes of RF and Cryo for AVNRT in the pediatric population. Methods AVNRT ablation outcomes were retrospectively analyzed utilizing the National Cardiovascular Data Registry IMPACT (Improving Pediatric and Adult Congenital Treatment) Registry from April 2016 to March 2019. Data from subjects 1 to 21 years of age undergoing elective first-time slow pathway (SP) modification for AVNRT were included. Exclusion criteria included <1 year of age, congenital heart disease, and >1 ablation target. Cases were analyzed by ablation energy: 1) RF only; 2) Cryo only; 3) radiofrequency ablation switching to cryoablation (RF→Cryo); and 4) cryoablation switching to radiofrequency ablation (Cryo→RF). The primary outcome was acute ablation failure. Secondary outcomes included in-hospital adverse events. Results Among 2,448 patients (mean age 13.6 ± 3.4 years, 60% female), RF only was employed in 43% (n = 1,046), Cryo only in 49% (n = 1,201), RF→Cryo in 6% (n = 135), and Cryo→RF in 66 (3%). Acute ablation failure occurred in 1.3% (n = 33), with no difference by energy source (1% in RF only, 1.5% in Cryo only, 1.5% in RF→Cryo, 3% in Cryo→RF; P = 0.5). Atrioventricular (AV) block requiring permanent pacemaker did not occur in any group; transient AV block occurred in 0.4% of the cohort, with no difference by group. Conclusions In this largest pediatric study of AVNRT ablation, RF and Cryo demonstrated comparable high acute success and rare documentation of AV block that did not result in temporary or permanent pacing. Longitudinal data are important for further comparison of these modalities with regard to recurrence risk and late complications.

ObjectivesTo describe the prevalence of Owlet Smart Sock (OSS) use in infants with supraventricular tachycardia (SVT) and associated demographic and clinical characteristics of users and to analyze the association of OSS use on medical resource use and clinical outcomes from emergency department (ED) encounters for SVT.Study designThis was a single-center, retrospective cohort study of infants with confirmed SVT from 2015 to 2022. OSS users and nonusers were compared across clinical and demographic parameters. Medical resource use (phone calls, office visits, ED visits) and outcomes (need for intensive care, length of stay, echocardiographic function, clinical appearance) were compared between OSS users and nonusers.ResultsOf 133 infants with SVT, OSS was used by 31 of 133 (23%), purchased before SVT diagnosis in 5 in 31 (16%) of users. No demographic difference was found between OSS users and nonusers. OSS users had more phone notes than nonusers, (P = .002) and more ED visits (P = .03), but the number of office visits and medication adjustments did not differ. During ED presentation, OSS users had better preserved left ventricular ejection fraction on echocardiogram (P = .04) and lower length of hospital stay by a mean 1.7 days (P = .02).ConclusionsOSS is used by a portion of infants with SVT. It is associated with more frequent phone calls and ED visits but lower length of stay and better-preserved cardiac function upon presentation.