{"slug": "echobridge-long-tail-aware-ecg-echocardiography-text-alignment-for-derived", "title": "EchoBridge: Long-Tail-Aware ECG-Echocardiography Text Alignment for Echocardiography-Derived Cardiac Findings", "summary": "EchoBridge, a new model for aligning ECG and echocardiography data, improves classifier-free AUROC, AUPRC, and F1 by 7.88, 5.61, and 4.54 points over baselines, according to a preprint on arXiv. The model uses Complementary Shared-Private Projection and Adaptive Prototype Boundary Calibration to handle long-tailed finding distributions and achieves top performance across multiple evaluation protocols.", "body_md": "arXiv:2607.24553v1 Announce Type: cross\nAbstract: Standardized echocardiography conclusions provide meaningful supervision for learning ECG representations of echocardiography-derived cardiac findings. Global ECG--text alignment may entangle modality-specific factors, while long-tailed finding distributions provide sparse positive supervision for low-prevalence conditions. We propose EchoBridge with Complementary Shared--Private Projection (CSPP) and Adaptive Prototype Boundary Calibration (APBC). CSPP maps each modality into shared and auxiliary private projections, reduces directional redundancy via within-modality orthogonality, and bidirectionally aligns normalized shared projections. APBC organizes the shared hypersphere with class-specific prototypes, training-frequency-adaptive angular margins, and spherical Riesz repulsion. We evaluate EchoBridge on EchoNext-Mini and independent PKUPH and SHTMU cohorts under four protocols: prompt-based inference without downstream classifier training, in-domain frozen linear probing, target-domain cross-center frozen linear probing, and source-only cross-center transfer, supplemented by finding-specific analyses. EchoBridge improves classifier-free AUROC, AUPRC, and F1 over the strongest baselines by 7.88, 5.61, and 4.54 points, respectively, and achieves the highest point estimates across all in-domain and target-domain probing budgets and both source-only transfer cohorts. Finding-specific analyses show gains for most conditions, including several low-prevalence valvular findings.", "url": "https://wpnews.pro/news/echobridge-long-tail-aware-ecg-echocardiography-text-alignment-for-derived", "canonical_source": "https://www.machinebrief.com/news/echobridge-long-tail-aware-ecg-echocardiography-text-alignme-t8wi", "published_at": "2026-07-28 04:00:00+00:00", "updated_at": "2026-07-28 04:57:23.481685+00:00", "lang": "en", "topics": ["machine-learning", "artificial-intelligence"], "entities": ["EchoBridge", "EchoNext-Mini", "PKUPH", "SHTMU", "arXiv"], "alternates": {"html": "https://wpnews.pro/news/echobridge-long-tail-aware-ecg-echocardiography-text-alignment-for-derived", "markdown": "https://wpnews.pro/news/echobridge-long-tail-aware-ecg-echocardiography-text-alignment-for-derived.md", "text": "https://wpnews.pro/news/echobridge-long-tail-aware-ecg-echocardiography-text-alignment-for-derived.txt", "jsonld": "https://wpnews.pro/news/echobridge-long-tail-aware-ecg-echocardiography-text-alignment-for-derived.jsonld"}}