CVAI Catalog

·

View Catalog

CVAI Catalog

Cardiovascular artificial intelligence models for discovery, comparison, and reuse

arrow_back
Return to Catalog

Website & database developed by Cam Ellinor

Catalog maintained by the Fahed Lab

CVAI Catalog © 2026 Broad Institute

PanEcho

Yale School of Medicine (CarDS Lab)

Echocardiography video

Filter catalog by Modality:
Echocardiography

LVEF estimation

Filter catalog by Disease / Trait:
Cardiac Function & Hemodynamics

LV dilation

Filter catalog by Disease / Trait:
Structural Heart & Cardiomyopathy

LV systolic dysfunction (LVSD)

Filter catalog by Disease / Trait:
Cardiac Function & Hemodynamics

Valvular disease

Filter catalog by Disease / Trait:
Structural Heart & Cardiomyopathy

Structural heart disease (composite)

Filter catalog by Disease / Trait:
Structural Heart & Cardiomyopathy

Regression

Filter catalog by Task Type:
Regression

Binary classification

Filter catalog by Task Type:
Classification

Multi-label classification

Filter catalog by Task Type:
Classification

Hybrid

Filter catalog by Architecture:
Hybrid / Multi-branch

PyTorch

Filter catalog by Framework:
PyTorch

CC BY-NC-SA 4.0

Filter catalog by License:
Non-commercial / Research-only

code View code & weights

View-agnostic, multi-task model that performs 39 different echocardiographic reporting tasks - covering chamber size and function, valve disease, and more - from any combination of views, aggregating clip-level predictions up to the study level. Combines a ConvNeXt-Tiny frame encoder with a temporal Transformer and separate output heads per task. Trained on private Yale-New Haven Health System echo videos and published in JAMA in 2025 by Yale's CarDS Lab.

memory Specifications

category

Architecture

Hybrid

ConvNeXt-Tiny frame encoder + temporal frame Transformer, with task-specific output heads for 39 tasks

code

Framework

PyTorch

calendar_month

Added to catalog

2026-07-10

gavel License

CC BY-NC-SA 4.0

check_small Open source close_small No commercial use check_small Modifications allowed Attribution required Share-alike required

License for model weights only. Associated code may be licensed seperately, check code source for specific terms.

description Publication

Complete AI-Enabled Echocardiography Interpretation With Multitask Deep Learning open_in_new

2025 · original paper

DOI: 10.1001/jama.2025.8731

database Training & evaluation data

RVENet+ Budapest Echocardiography External Validation Cohort (PanEcho)

test

public 831 subjects · Hungary

Yale New Haven Health POCUS ED External Validation Cohort (PanEcho)

test

public 3,170 subjects · USA

Yale New Haven Health System Echocardiography Cohort (PanEcho)

train

public 24,405 subjects · USA · 2016-2022

24,405 unique patients (32,265 studies); development split 18,343 patients, internal validation split 4,588 patients

science Capabilities & performance

LVEF (study-level)

Regression LVEF estimation
4.4 MAE YNHHS internal test (temporally distinct) · internal
5.5 MAE External geographically distinct cohorts · external

Moderate-or-greater LV dilation

Binary classification LV dilation
0.95 AUROC YNHHS internal test (temporally distinct) · internal
0.98 AUROC External geographically distinct cohorts · external

LV systolic dysfunction

Binary classification LV systolic dysfunction (LVSD)
0.98 AUROC YNHHS internal test (temporally distinct) · internal
0.94 AUROC External geographically distinct cohorts · external

Severe aortic stenosis

Binary classification Valvular disease
0.99 AUROC YNHHS

Multi-label classification

Structural heart disease (composite)