BACKGROUND:Pre-participation screening (PPS) in competitive athletes aims to identify cardiovascular diseases associated with sudden cardiac death (SCD). Although the 12‑lead electrocardiogram (ECG) represents the cornerstone of PPS, structural abnormalities may demonstrate limited or incomplete electrical expression, particularly in asymptomatic athletes with physiological remodeling. Artificial intelligence (AI)-enabled ECG models have shown promising performance in hospital-based populations, but their transportability to low-prevalence athlete screening environments remains uncertain. OBJECTIVES:To develop and externally validate a deep learning (DL)-based AI-ECG ensemble model for detecting imaging-confirmed structural heart disease in competitive athletes undergoing PPS. METHODS:A convolutional neural network (CNN) ensemble was trained using hospital-derived ECG images from Beth Israel Deaconess Medical Center (BIDMC, Boston, USA) and externally validated in the Italian Team for Athlete CARDiac evaluation and AI-based Risk prediction (ITACARD-AI) registry. Separate CNNs were developed for valvular heart disease (VHD) and cardiomyopathies (CM) and combined using XGBoost meta-learning. Model performance was assessed using area under the receiver operating characteristic curve (AUROC), subgroup analyses, and threshold-based evaluation. RESULTS:The ITACARD-AI cohort included 1115 competitive athletes (mean age 26 ± 13 years; 70% male), including 48 athletes (4.3%) with VHD and 30 (2.7%) with CM. External validation demonstrated substantial performance degradation compared with hospital-based internal validation. AUROC values decreased to 0.70 (95% CI 0.64-0.75) for VHD and 0.69 (95% CI 0.60-0.78) for CM, indicating only modest discrimination in the screening population. Threshold analyses showed high negative predictive values (∼99%) but persistently low positive predictive values (≤8%), reflecting limited disease enrichment and strong prevalence dependence. CONCLUSIONS:Hospital-trained AI-ECG models demonstrated limited transportability to real-world athlete screening populations. The marked reduction in external performance suggests that low disease prevalence, heterogeneous physiological remodeling, and incomplete ECG expression of structural abnormalities may substantially limit class separability in PPS environments. Although AI-ECG may provide cautious adjunctive support within physician-led workflows, these findings highlight the intrinsic challenges of applying ECG-based AI to low-prevalence sports cardiology screening populations.

Deep neural networks can classify ECGs with high accuracy when training data is abundant. Rare conditions like Brugada syndrome, an inherited arrhythmia syndrome predisposing to sudden death, pose challenges due to data scarcity hindering model training. We evaluated multiple machine learning (ML) approaches to optimise a Brugada ECG classification model using limited training data. The baseline model was trained on a dataset comprising 176 Brugada, 176 right bundle branch block (RBBB) and 352 normal ECGs from Zhongshan Hospital (Zhongshan-baseline dataset), framed as a binary classification task to distinguish Brugada from non-Brugada ECGs. A 25%-75% train-test split was used to exacerbate data scarcity. To enhance training, we incorporated three additional datasets: (i) a different, labelled ECG dataset from Zhongshan Hospital including normal and RBBB ECGs (Zhongshan-pretrain), (ii) an unlabelled ECG dataset from Hammersmith Hospital including Brugada and non-Brugada ECGs (Imperial), (iii) an open-access labelled ECG dataset (PTB-XL). Three strategies were tested: (1) supervised pretraining, (2) self-supervised pretraining with data augmentation, and (3) oversampling using SMOTE (synthetic minority oversampling technique). Each model was evaluated on the unseen internal test set and an external Brugada mimic dataset. The models were re-trained using an 80%-20% train-test split as a secondary analysis. The baseline model achieved 92.2% accuracy, F1-score 0.837, and area under the Receiver Operating Characteristic curve (AUC) 0.962. Supervised pretraining significantly improved performance when training data was scarce, with the best model pretrained on the Zhongshan-pretrain dataset boosting accuracy (+3.2%), F1-score (+0.071) and AUC + 0.019), with consistent cross-validation performance. Self-supervised pretraining produced smaller and more variable gains, although select models better mitigated against false positives on the Brugada mimic dataset. SMOTE oversampling showed inconsistent effects on performance. Incorporating pretraining and oversampling may facilitate the development of more accurate AI-ECG models for rare diseases when training data is limited but provides diminishing returns when adequate labelled data is available.

Contrastive learning is a widely adopted self-supervised pretraining strategy, yet its dependence on cohort composition remains underexplored. We present Contrasting by Augmented Patient Electrocardiograms (CAPE) foundation model and pretrain on four cohorts (n = 5,203,269), from diverse populations across three continents (North America, South America, Asia). We systematically assess how cohort demographics, health status, and population diversity influence the downstream performance for prediction tasks also including two additional cohorts from another continent (Europe). We find that downstream performance depends on the distributional properties of the pretraining cohort, including demographics and health status. Moreover, while pretraining with a multi-centre, demographically diverse cohort improves in-distribution accuracy, it reduces out-of-distribution (OOD) generalisation of our contrastive approach by encoding cohort-specific artifacts. To address this, we propose the In-Distribution Batch (IDB) strategy, which preserves intra-cohort consistency during pretraining, discourages learning of spurious cohort-specific features, and instead promotes clinically meaningful variability within cohorts. This leads to improved out-of-distribution robustness, with gains of 9-40% in downstream label prediction performance. This work provides insights into pretraining strategies for more clinically deployable and generalisable foundation models.

Aims:A significant proportion of type 2 diabetes cases remain undiagnosed despite screening advances, carrying substantial cardiometabolic risk. Artificial intelligence-enhanced electrocardiography (AI-ECG) detects subtle ECG changes in subclinical disease, potentially enabling opportunistic screening. Methods and results:We developed AI-ECG Risk Estimator for Diabetes Mellitus (AIRE-DM), a convolutional neural network with discrete-time survival loss, for diagnosis of prevalent and prediction of incident type 2 diabetes. It was trained on 1 163 401 ECGs from 189 537 individuals from Beth Israel Deaconess Medical Center (BIDMC) and externally validated in UK Biobank (UKB; n = 65 606) and ELSA-Brasil (n = 13 739). AI-ECG Risk Estimator for Diabetes Mellitus demonstrated moderate discrimination for prevalent type 2 diabetes (area under the receiver operating characteristic curve: BIDMC 0.724, UKB 0.733, ELSA-Brasil 0.706) and incident type 2 diabetes (C-index: BIDMC 0.667, UKB 0.688, ELSA-Brasil 0.625). The highest AIRE-DM risk quartile had elevated incident diabetes risk vs. the lowest (hazard ratio: BIDMC 4.75, UKB 7.52, ELSA-Brasil 3.96). AI-ECG Risk Estimator for Diabetes Mellitus was non-inferior to the American Diabetes Association Diabetes Risk Test in BIDMC, with improved predictive accuracy when combined. In normoglycaemic patients, AIRE-DM was superior to glycated haemoglobin (HbA1c) for predicting incident diabetes in BIDMC and non-inferior in ELSA-Brasil. The highest risk quartile reached 5% cumulative type 2 diabetes mellitus incidence 5.4 years (BIDMC) and 4.8 years (ELSA-Brasil) earlier than the lowest risk quartile, after adjusting for HbA1c, age, and sex. Phenome- and genome-wide association studies revealed biologically plausible associations with glucose regulation, cardiac morphology, diastolic dysfunction, arterial stiffness, and lipid metabolism. Conclusion:AI-ECG Risk Estimator for Diabetes Mellitus detects prevalent type 2 diabetes and predicts incident disease, uniquely identifying high-risk individuals within the normoglycaemic range. Combined with clinical scores or biomarkers, it enhances risk stratification, enabling earlier intervention.

Right ventricular (RV) trabeculation may reflect adaptation to loading conditions and imprinting of early developmental processes, yet its genetic determinants and clinical relevance in adults remain poorly understood. In contrast to the left ventricle, the RV has distinct developmental origins, geometry and loading conditions, suggesting chamber-specific mechanisms of trabecular remodelling. Using deep learning-based image segmentation and fractal dimension analysis, we quantified RV trabecular complexity in diastole and systole in 48,118 UK Biobank participants with genetic data. RV trabecular morphology was associated with systolic function, as well as cardiometabolic and respiratory conditions. Genetic analyses across the allele frequency spectrum identified 52 common loci and 45 genes with a burden of protein-altering variants, implicating sarcomeric function, cytoskeletal organisation, and early cardiac patterning. Although many loci showed shared effects across both ventricles, we also identified RV-specific genetic associations linked to congenital heart disease and respiratory phenotypes. Notably, associations at CFTR suggest a connection between airway-associated mucus regulation and RV remodelling, whereas MYH6 implicates sarcomeric and developmental mechanisms in adult RV patterning. Together, these findings establish RV trabeculation as a trait that captures both shared and chamber-specific mechanisms relevant to cardiovascular health and disease.

Background:Differentiating chronic from acute left bundle branch block (LBBB) is useful in various cardiac disorders. Peak QRS/T ratio, a measure of cardiac memory, can differentiate chronic from acute LBBB with high sensitivity and specificity, but its utility in post-transcatheter aortic valve replacement (TAVR) patients with LBBB is unclear. Objective:This study aimed to validate the QRS/T ratio for differentiating chronic/acute LBBB in post-TAVR patients and assess whether the spatial ventricular gradient (SVG), which integrates 3-dimensional depolarization/repolarization throughout the cardiac cycle, could also distinguish chronic from acute LBBB. Methods:This was a post hoc analysis of a prospective, observational, single-center TAVR study. After TAVR, patients were classified as acute (new <24 hours) or chronic/preexisting LBBB. Electrocardiograms were transformed into vectorcardiograms, and peak QRS/T ratio and SVG were calculated. Sensitivity, specificity, and area under the receiver-operating curve (AUROC) assessed how QRS/T ratio and SVG differentiated chronic from acute LBBB. Results:Of 409 patients, 21 had preexisting, and 53 had acute/new LBBB. Patients with chronic LBBB had a higher peak QRS/T ratio than patients with acute LBBB (median 3.8 vs 2.2; P < .0001). 3-dimensional SVG vectors significantly differed between chronic and acute LBBB (joint P < .0001), due to differences in the anterior/posterior (Z) direction, (SVGz): median 43.4 vs -14.2 mV·ms, P < .0001, for chronic vs acute LBBB, respectively. A peak QRS/T ratio of ≥2.7 and an SVGz of ≥15 mV·ms had 100% sensitivity, 98.1% specificity, and an AUROC of 0.991 for chronic LBBB. Over long-term follow-up, patients transitioned from acute to chronic LBBB vectorcardiographic phenotype. Conclusion:Peak QRS/T ratio and SVG differentiate chronic from acute LBBB in post-TAVR patients with 100% sensitivity, 98% specificity, and an AUROC of 0.99.

Cardiac MRI (CMR) markers of myocardial fibrosis and infiltration are diagnostically and prognostically important in cardiomyopathies. A noninvasive ECG correlate of CMR interstitial fibrosis measurements (extracellular volume [ECV] and native T1) could assist in diagnosis, risk stratification, and tracking disease progress. The spatial ventricular gradient (SVG) is a vectorcardiographic (VCG) measure of electrical heterogeneity obtained from a 12-lead ECG. The link between the SVG and CMR-derived myocardial interstitial fibrosis is unknown. Retrospective study of patients referred for CMR from 2018–2022 at a single academic center, with an ECG performed within 30 days. VCGs were constructed from 12-lead ECGs, and SVG vector components were calculated. ECV and T1 values were regressed on SVG components, demographics, and ECG parameters using linear regression. In total, 345 patients met inclusion criteria: 55

Background: Adjunctive posterior wall isolation (PWI) to pulmonary vein isolation (PVI) has not demonstrated convincing benefit during atrial fibrillation (AF) ablation. To provide mechanistic insight for null PWI trials, we undertook Granger causality (GC) analysis of noncontact left atrial (LA) electroanatomic maps. Objective: This study aimed to apply GC to intracardiac electrograms to uncover patient-specific AF dynamics and describe a proof-of-concept approach to targeted PWI after PVI. Methods: A prospective cohort study was undertaken at Royal Brompton Hospital. Consecutive patients undergoing PVI with noncontact mapping (AcQmap; Acutus Medical) before and after PVI were included. Results: In 21 patients, causality pairing index, a GC measure of organization, was unchanged after PVI (overall, 0.087 ± 0.012 vs 0.086 ± 0.015; P = .64) or by region (posterior wall [PW], 0.084 ± 0.020 vs 0.079 ± 0.017; P = .20; rest of LA, 0.087 ± 0.013 vs 0.086 ± 0.016; P = .80). Directional dispersion, quantifying conduction heterogeneity, was lower in the PW than the rest of the LA (0.093 ± 0.036 vs 0.11 ± 0.043; P = .017) and increased after PVI (0.093 ± 0.036 vs 0.12 ± 0.043; P = .045), whereas there was no change in the rest of the LA (0.11 ± 0.034 vs 0.11 ± 0.030; P = .52). PW net outflow overall decreased after PVI (before, −0.0086 ± 0.047 vs −0.033 ± 0.054; P = .011) with a minority of patients exhibiting a net positive outflow from the PW. Conclusion: GC provides mechanistic insight into the null trials for PWI and identifies a minority of patients who may benefit. GC is positioned as a clinical decision tool to guide personalized persistent AF ablation strategies.

BACKGROUND Early prediction of atrial fibrillation (AF) is crucial for reducing adverse outcomes. While artificial intelligence-enhanced electrocardiogram (AI-ECG) analysis shows promise in predicting AF, most approaches require digital ECG signals, limiting their application in settings where ECGs are stored as images. OBJECTIVE We aimed to develop and validate an image-based AI-ECG approach for predicting incident AF across multiple datasets. METHODS We used 1,163,401 ECGs from 189,539 patients in the Beth Israel Deaconess Medical Center(BIDMC) dataset and 70,655 ECGs from 65,610 participants in the United Kingdom (UK) Biobank. The AI-ECG model was trained on ECG images processed to 310x868 pixels. RESULTS The model achieved C-statistics of 0.754 (95% confidence interval [CI]: 0.747-0.761) in the BIDMC dataset and 0.723 (95% CI: 0.704-0.741) in the UK Biobank for predicting incident AF. Performance was maintained across key subgroups including outpatients, women, and non-white individuals. Compared with the CHARGE-AF risk score, the AI-ECG model showed superior performance (c-statistic 0.696 vs 0.667, P < .05) and provided significant additive value when combined (c-statistic 0.711, P < .0001). The model also performed well on smartphone-photographed ECGs (c-statistic 0.736). Saliency mapping indicated the model primarily focused on P-wave morphology and PR interval regions. CONCLUSION This image-based approach enables AI-ECG prediction of AF in settings without digital ECG infrastructure and provides additive value to known clinical risk scores