BACKGROUND AND AIMS:The 12-lead electrocardiogram (ECG) remains a cornerstone of cardiac diagnostics, yet existing artificial intelligence (AI) solutions for automated interpretation often lack generalizability, remain closed source, and are primarily trained using supervised learning (SL), which requires extensive labelled datasets and may limit adaptability across diverse clinical settings. Self-supervised learning (SSL) can potentially overcome these limitations by learning robust representations from unlabelled data. To address these challenges, this study developed and compared two open-source foundational ECG models: DeepECG-SL, a supervised multilabel ECG model, and DeepECG-SSL, a self-supervised model. METHODS:Both models were trained on over 1 million ECGs using a standardized preprocessing pipeline and automated free-text extraction from ECG reports to predict 77 cardiac conditions. DeepECG-SSL leveraged unlabelled data through self-supervised contrastive learning and masked lead modelling before fine-tuning for downstream tasks, while DeepECG-SL was trained directly on labelled diagnostic data in an end-to-end fashion. Performance was evaluated across seven private, multilingual healthcare systems and four public ECG repositories, with assessment of fairness by age and sex, and investigation of privacy vulnerabilities as well as memory and compute requirements. RESULTS:DeepECG-SSL achieved micro-averaged area under the receiver operating characteristic curves (AUROCs) across all 77 cardiac conditions for ECG interpretation of 0.990 [95% confidence interval (CI): 0.990, 0.990] on the internal dataset (MHI-ds), 0.981 (95% CI: 0.981, 0.981) on external public datasets (UKB, CLSA, MIMIC-IV and PTB), and 0.983 (95% CI: 0.983, 0.983) on external private datasets (UW, UCSF, JGH, NYP, MGH, CSH and CHUM), while DeepECG-SL demonstrated AUROCs of 0.992 (95% CI: 0.992, 0.992), 0.980 (95% CI: 0.980, 0.980), and 0.983 (95% CI: 0.983, 0.984), respectively. Fairness analyses revealed minimal disparities (true-positive rate and false-positive rate difference <0.1) across age and sex groups for both models. DeepECG-SSL demonstrated superior performance on limited-data digital biomarker tasks, with the largest improvements in long QT syndrome (LQTS) genotype classification (AUROC 0.931 vs 0.850, P = .026, n = 127 ECGs) and 5 year atrial fibrillation risk prediction (AUROC 0.742 vs 0.734, P < 0.001, n = 132 050 ECGs), while achieving superior performance in left ventricular ejection fraction ≤40% classification (AUROC 0.926 vs 0.917, P < 0.001, n = 25 252 ECGs) and comparable performance in LQTS detection (AUROC 0.767 vs 0.735, P = 0.117, n = 934 ECGs). CONCLUSIONS:This study establishes SSL as a promising paradigm for ECG analysis, particularly in settings with limited annotated data, enhancing accessibility, generalizability, and fairness in AI-driven cardiac diagnostics. By releasing model weights, preprocessing tools, and validation code, this work aims to support robust, data-efficient AI diagnostics across diverse clinical environments and questions.
BACKGROUND:Cardiomyocytes, as highly specialized and differentiated somatic cells, possess a limited capacity for renewal. Neonatal rodents possess the ability to regenerate cardiomyocytes after injury; however, this regenerative capacity declines rapidly with cardiomyocyte maturation, suggesting an inhibitory network between cellular maturation and cardiomyocyte proliferation. Maturing cardiomyocytes undergo a metabolic shift from predominantly glycolysis in the neonatal state to increased fatty acid oxidation in the mature state, which poses a barrier to cardiomyocyte proliferation and cardiac regenerative repair. YAP, a transcriptional cofactor regulated by the Hippo signaling pathway, promotes cardiac regenerative repair. We investigated the role of YAP in mediating metabolic remodeling to overcome the cardiomyocyte proliferation barrier and enable cardiac regenerative repair after heart injury. METHODS:We explored how YAP induces metabolic remodeling through single-nucleus RNA sequencing and metabolomic analyses in mice. Using lipidomic analysis, we demonstrated how YAP remodels the balance of fatty acid catabolism and anabolism. We further used a maternal fat overloading model to stimulate fatty acid oxidation, which activates a maturation program in neonatal cardiomyocytes and counteracts YAP-mediated metabolic dematuration. Using chromatin accessibility (assay for transposase-accessible chromatin with high-throughput sequencing), DNA footprinting, and transcriptional profiling (RNA sequencing), we discovered the key transcription factors that YAP interrupts to reprogram the cardiomyocyte metabolic state. RESULTS:Our results demonstrate that YAP directs metabolic remodeling of mature cardiomyocytes toward a neonatal-like metabolic state and illustrate the role of fatty acid metabolism in proliferating cardiomyocytes. We found that YAP reduces cardiomyocyte fatty acid utilization, driving fatty acid anabolism and phospholipid biosynthesis. Genome-wide analyses revealed that YAP inhibits the cardiac maturation transcription factor MEF2A (myocyte-specific enhancer factor 2A), resulting in decreased gene expression of cardiomyocyte maturity pathways. Given the role of MEF2A in regulating contractility, energy production, and mitochondrial homeostasis, we found that perturbing MEF2A transcriptional activity can serve as a strategy to interrupt the cardiomyocyte maturation program and restore the regenerative capacity of the heart. CONCLUSIONS:Our research endeavors to provide a comprehensive understanding of the balance of cardiomyocyte metabolic maturation and proliferation to overcome barriers to heart regeneration, offering novel insights into the potential for therapeutic intervention in heart failure.
Wireless communication technologies for bioelectronic implants enable remote monitoring for diagnosis and adaptive therapeutic intervention without the constraints of wired connections. However, wireless data uplink from millimeter-scale devices deep in the body struggles to achieve low power consumption while maintaining large misalignment tolerances. Here, we report a passive wireless backscatter communication system based on magnetoelectric transducers that consumes less than 0.3 pJ/bit and achieves less than 1E-6 bit error rate at a distance of 55 mm while tolerating a misalignment of 10 mm. Using this robust data uplink, we designed a wireless cardiac sensing node that can transmit electrocardiogram signals from the beating heart surface of a porcine model to a custom external transceiver using the magnetoelectric backscatter uplink. This reliable, near-zero-power communication method provides opportunities for next-generation bioelectronics to feature real-time physiological monitoring and closed-loop therapies while maintaining a small form factor and low power consumption.
Background Left ventricular (LV) thrombus is typically associated with systolic dysfunction. Its occurrence after heart transplantation (HTx) with preserved ejection fraction (EF) is rare. Case Summary A 58-year-old man, 3 years post-HTx, was incidentally found to have an LV apical mass on routine echocardiography with preserved EF. Cardiac magnetic resonance and computed tomography suggested thrombus. The lesion persisted after direct oral anticoagulant therapy but resolved after transition to a vitamin K antagonist. Retrospective review revealed an apical false tendon, likely serving as a nidus for thrombus formation. Discussion This case highlights LV thrombus formation post-HTx despite normal EF with no clinical evidence of rejection, possibly related to subtle structural anomalies. Multimodality imaging was essential for differentiating thrombus from tumor and guiding optimal anticoagulation therapy. Take-Home Messages Subtle LV structural variants may predispose to thrombus despite normal EF. Vitamin K antagonists may be more effective than direct oral anticoagulant in selected cases, underscoring its value in evaluating cardiac masses in complex posttransplant patients.
Objectives/Goals: Blood pressure (BP) control remains suboptimal among US patients with hypertension. Single-pill combination (SPC) therapies are commonly used to improve adherence; however, their effectiveness for achieving early and sustained intensive BP control is unclear. Methods/Study Population: We performed a post hoc analysis of SPRINT including 2,736 participants propensity matched in 1:2 ratio to compare effects of SPCs with equivalent multi-pill therapy. The estimated marginal odds of achieving optimal BP control were derived using generalized linear mixed models with repeated measures (LMMRM). The association between time-updated SPC use and BP change in short- (≤6 months) and long-term (>6 months) follow-up was assessed with LMMRM and SPC*time interaction term. Multivariable Cox models evaluated association of SPC use with CV events and serious adverse events (SAEs). Results/Anticipated Results: Among SPRINT participants (N=8623), 9.3% (N=803) were prescribed SPC at baseline with greater use in the intensive vs. usual care group (5.79 vs. 3.90 per 100 person-months; p-diff<0.001). Among matched pairs (SPC[n=912); multi-pill therapy[n=1824]), SPC use was associated with 22% increased likelihood of achieving target BP by 6 months [OR(95% CI): 1.22(1.05, 1.42)]. Participants receiving SPCs (vs multi-pills) experienced more rapid BP reduction in the first 6 months (-2.0 vs. -1.2 mmHg monthly change; p-diff<0.001). Over long-term follow-up, participants using SPCs achieved significantly lower SBP at each timepoint. The risk of the primary CV composite endpoint and SAEs was not significantly different between groups. Discussion/Significance of Impact: SPC therapy resulted in more rapid and sustained BP reduction and a greater likelihood of achieving BP control compared with multi-pill therapy without increase in SAEs. Future research is needed to identify optimal strategies for implementing SPC-based approaches at the population-level and optimize public health benefits of intensive BP control.
