ecg

DCRA introduces a diffusion-conditioned representation alignment framework to enhance robustness in time-series learning, particularly for clinical signals like EEG and ECG, by aligning representations across noise levels.

This paper introduces DCGCNet, a novel deep learning framework for accurate and robust atrial fibrillation detection from ECG signals, demonstrating state-of-the-art performance and strong generalization across diverse datasets and noisy conditions.

The paper proposes P2E-VQ, a retrieval-augmented framework that enhances PPG representations by retrieving ECG-linked information via discrete patch retrieval, improving downstream tasks without requiring ECG during inference.

Introduces CardioState-JEPA, a cardiac foundation model that learns a shared representation across ECG, PPG, and PCG signals using a delay-aware joint-embedding predictive architecture, improving downstream cardiac classification tasks.

This paper presents a controlled study on ECG self-supervised representation learning, examining how temporal context length (16s to 10min) and encoding strategy (continuous patch embeddings vs discretized tokens) affect downstream rhythm detection and patient-level retrieval. Results show longer context and continuous encoders improve performance, motivating extended-context ECG foundation models.

This paper presents a model-agnostic framework for per-modality failure analysis in multimodal clinical AI, distinguishing loud vs silent failures when a modality is dropped. Validated on planted ground truth and applied to EchoJEPA and HuBERT-ECG embeddings for LVEF prediction, it shows that dropping echo nearly doubles error.

ECG-InterpBench is a new benchmark that systematically evaluates the interpretability of ECG foundation model representations using matched-scale sparse autoencoders, covering reconstruction fidelity, clinical concept accessibility, and reproducibility across 450 cells.

CADENCE uses sparse autoencoders to decompose ECG foundation model representations into interpretable physiological concepts, significantly improving alignment with clinical phenotypes and waveform morphology.

Proposes PEACE, a knowledge-guided framework for transferring adult ECG interpretation to pediatric populations using label-conditioned contrastive alignment, achieving significant improvements under limited supervision.

This study evaluates nine ECG foundation models for Brugada syndrome detection, finding that pre-training provides optimization stability but not transferable clinical knowledge, challenging assumptions about the benefits of large-scale pre-training for rare diseases.

This paper introduces IRFE-ECG, a method for continual ECG deployment that separates expert retention from autonomous source inference using frozen features from ECGFounder, achieving strong performance without replaying raw ECGs.

The paper proposes using spectral entropy as a metric to quantify noise introduced by explainability techniques in ECG arrhythmia classification, helping to distinguish true model signal from XAI-generated artifacts.

ATRIA is a multi-agent system for ECG report generation that mirrors the clinician's iterative workflow, enabling bidirectional editing, evidence grounding, and clinician-in-the-loop verification.

Proposes MSAIC-Net, a multi-scale attention-enhanced convolutional network for detecting myocardial substrate abnormalities from ECG signals, using imbalance-aware contrastive learning and lead-wise permutation importance for interpretability.

ADAPTOOD is a novel framework that uses data uncertainty to quantify distribution shift severity and guide fine-tuning of ECG time series models for out-of-distribution settings. It combines uncertainty estimation with low-rank model updates and adaptive hyperparameter optimization, achieving up to 7% higher accuracy and 12.9% higher precision than existing OOD adaptation methods.

This paper introduces StenCE, a pretraining framework that uses cross-modal contrastive learning between ECG and X-ray angiography representations to detect severe coronary stenosis from ECGs, achieving high performance and enabling early diagnosis even in asymptomatic patients.

DeepArrhythmia is a multimodal framework for beat-level ECG arrhythmia classification that combines raw ECG signals and waveform images, using segment-level confidence to selectively acquire physiological evidence for improved accuracy.

MIT researchers have developed PULSE-HF, a deep learning model that predicts whether heart failure patients will experience worsening left ventricular ejection fraction within a year using electrocardiograms. The model, published in Lancet eClinical Medicine, could help clinicians prioritize high-risk patients and reduce unnecessary hospital visits in both well-resourced and low-resource clinical settings.