PanEcho

PanEcho is a multitask deep learning system developed to automate comprehensive interpretation of transthoracic echocardiograms, aiming to reduce reliance on manual expert reporting. Trained and validated on over 1.2 million echocardiographic videos from more than 24,000 patients across multiple internal and external cohorts, the model achieved high accuracy across 39 diagnostic and measurement tasks, including precise estimation of left ventricular ejection fraction and detection of major cardiac pathologies. PanEcho maintained strong performance even on limited and point-of-care ultrasound studies, supporting its potential use as an adjunct reader in echo labs and as a screening tool in resource-constrained or rapid-care settings.