Research

The papers came first.

InVision's products are not inspired by the literature — they are the literature, cleared. Every model in the platform was developed, peer-reviewed, and published before it became a product, and several were released as open datasets and open weights so other groups could test them without our permission.

Every paper, year by year

56 echocardiography publications in total. The 16 below are the ones the platform is built on — including three in Nature, and one independent evaluation run by a group outside InVision. The full list follows.

2020

EchoNet-Dynamic · Nature

The first video-based deep learning model for beat-to-beat ejection fraction. The foundation the platform is built on.

Multi-task echo interpretation · npj Digital Medicine

Identifies cardiac structures, estimates volumetric measurements and function, and predicts systemic phenotypes — with the first interpretation framework applied to echo deep learning.

Cardiac function
2021

EchoNet-Labs · EBioMedicine

Laboratory values predicted directly from echo video, showing the study carries signal well beyond the measurements written into the report.

Opportunistic detection
2022

EchoNet-LVH · JAMA Cardiology

Development of the model now cleared as InVision Precision Cardiac Amyloid. 23,745 patients; amyloidosis AUC 0.83, HCM AUC 0.98.

Cardiomyopathy & amyloidosis
2023

EchoNet-RCT · Nature

The first blinded, randomized clinical trial of AI in cardiology. 16.8% vs 27.2% substantial change; NCT05140642.

Clinical trial
2024

EchoNet-MR · Circulation

Mitral regurgitation severity assessed directly from color Doppler video across 58,614 studies, with external validation at a geographically distinct health system.

Cohort construction · JACC: Advances

Model AUC moved from 0.660 to 0.898 on cohort construction alone. The reason cross-study AUC comparisons are not valid.

EchoCLIP · Nature Medicine

Vision-language pretraining on echo studies and their reports.

Occult atrial fibrillation · npj Digital Medicine

A two-stage video model that flags patients in sinus rhythm who had atrial fibrillation within 90 days — disease the study itself does not show.

Valvular & pericardial · Cardiomyopathy & amyloidosis · Foundation models · Opportunistic detection
2025

Independent head-to-head · JACC: Advances

Northwestern-led head-to-head, run without InVision's involvement, including a fairness audit. We publish the whole table.

Precision Cardiac Amyloid · JACC: Advances

Five sites across the US and Japan. AUC 0.893; PPV 0.954. Site-level results published, including the lowest.

Tricuspid regurgitation · JAMA Cardiology

TR severity from color Doppler video. Moderate-or-severe AUC 0.928; severe AUC 0.956.

EchoNet-Measurements · Journal of the American College of Cardiology

877,983 annotations across 155,215 studies. 18 parameters at R² 0.967–0.987.

EchoNet-Liver · NEJM AI

Cirrhosis and steatotic liver disease from the subcostal view already captured in a routine echo. 1.5M+ videos.

Cardiac age · npj Digital Medicine

Predicted cardiac age from 2,610,266 videos across 166,508 studies, mean absolute error 6.76 years — a marker for cardiovascular disease independent of chronological age.

Cardiomyopathy & amyloidosis · Valvular & pericardial · Measurement & quality · Opportunistic detection
2026

EchoPrime · Nature

12M+ video-report pairs, state of the art on 23 benchmarks, validated across five health systems.

Foundation models

Research threads

The work runs along several parallel lines. Each product traces back to one of them, and the methods work underneath applies to all of them.

Cardiac function

EchoNet-Dynamic · Multi-task echo interpretation

16 publications

Cardiomyopathy & amyloidosis

EchoNet-LVH · Cohort construction · Independent head-to-head

11 publications

Clinical trial

EchoNet-RCT

1 publication

Foundation models

EchoCLIP · EchoPrime

3 publications

Measurement & quality

EchoNet-Measurements

6 publications

Opportunistic detection

EchoNet-Labs · Occult atrial fibrillation · EchoNet-Liver

9 publications

Valvular & pericardial

EchoNet-MR · Tricuspid regurgitation

10 publications

All publications

Every echocardiography publication, newest first — journal articles, conference papers, abstracts, datasets, and preprints. Entries with a resolvable link are linked; the rest are listed in full so the citation can be looked up.

Year Journal Paper Type
2026 medRxiv Artificial intelligence-enabled echocardiographic assessment of right ventricular function Right ventricular function · M Tokodi, B He, Á Szijártó, A Ferencz, K Shiida, M Tolvaj, A Fábián, ... Preprint
2026 Pediatric Cardiology Predicting cardiac magnetic resonance-derived ejection fraction from echocardiogram via deep learning approach in Tetralogy of Fallot Echo-to-CMR ejection fraction · A Adhikari, GV Wesley III, MB Nguyen, TT Doan, MY Rao, A Parthiban, ... Journal
2026 Nature Comprehensive echocardiogram evaluation with view primed vision language AI ↗ EchoPrime · M Vukadinovic, IM Chiu, X Tang, N Yuan, TY Chen, P Cheng, D Li, ... Journal
2026 Journal of the American Society of Echocardiography A Clinically Interpretable AI System for Real-Time Quality Control of Transthoracic Echocardiography: Development, Validation, and Deployment Real-time quality control · Z Shi, H Cheng, Z Qi, C Shan, CC Taub, D Ouyang, C Lee, R Chen, Y Du, ... Journal
2026 JACC: Advances Detection of left ventricular outflow obstruction from standard B-mode echocardiogram videos using deep learning LV outflow obstruction · V Yuan, H Ieki, C Binder, Y Sahashi, PC Cheng, D Ouyang Journal
2025 npj Digital Medicine Artificial intelligence prediction of age from echocardiography as a marker for cardiovascular disease ↗ Cardiac age · M Rawlani, H Ieki, C Binder, V Yuan, IM Chiu, A Bhatt, JE Ebinger, ... Journal
2025 medRxiv Automated deep learning pipeline for characterizing left ventricular diastolic function Diastolic function · V Yuan, Y Sahashi, H Ieki, M Vukadinovic, C Binder, K Pieszko, ... Preprint
2025 medRxiv Comprehensive aortic stenosis characterization using multi-view deep learning Aortic stenosis characterisation · H Ieki, Y Sahashi, M Vukadinovic, M Rawlani, C Binder, N Yuan, ... Preprint
2025 medRxiv Automated aortic regurgitation detection and quantification: A deep learning approach using multi-view echocardiography ↗ Aortic regurgitation · C Binder, Y Sahashi, H Ieki, M Vukadinovic, V Yuan, M Rawlani, P Cheng, ... Preprint
2025 PhysioNet Mimic-iv-echo-ext-mimicechoqa: A benchmark dataset for echocardiogram-based visual question answering ↗ Echo question answering · R Thapa, A Li, Q Wu, B He, Y Sahashi, C Binder-Rodriguez, A Zhang, ... Dataset
2025 Open Heart Ratio of interventricular septal thickness to global longitudinal strain accurately identifies cardiac amyloidosis Septal thickness to strain ratio · L Cao, GJ Hong, M Abiragi, J Le, PR Tacon, IM Chiu, J Patel, LK Stern, ... Journal
2025 NEJM AI Opportunistic screening of chronic liver disease with deep-learning–enhanced echocardiography ↗ EchoNet-Liver · Y Sahashi, M Vukadinovic, F Amrollahi, H Trivedi, J Rhee, J Chen, ... Journal
2025 Journal of the American Society of Echocardiography Using deep learning to predict cardiovascular magnetic resonance findings from echocardiographic videos Echo-to-CMR prediction · Y Sahashi, M Vukadinovic, G Duffy, D Li, S Cheng, DS Berman, D Ouyang, ... Journal
2025 Journal of the American Heart Association Understanding transient left ventricular ejection fraction reduction during atrial fibrillation with artificial intelligence ↗ EF during atrial fibrillation · N Yuan, GJ Hong, A Vrudhula, AC Kwan, G Duffy, P Botting, SS Dhruva, ... Journal
2025 Journal of the American College of Cardiology Artificial intelligence automation of echocardiographic measurements ↗ EchoNet-Measurements · Y Sahashi, H Ieki, V Yuan, M Christensen, M Vukadinovic, ... Journal
2025 Journal of Cardiology AI-echocardiography: Current status and future direction Field review · Y Sahashi, D Ouyang, H Okura, N Kagiyama Journal
2025 JAMA Cardiology Automated deep learning phenotyping of tricuspid regurgitation in echocardiography ↗ Tricuspid regurgitation · A Vrudhula, M Vukadinovic, C Haeffele, AC Kwan, D Berman, D Liang, ... Journal
2025 JACC: Advances International validation of echocardiographic artificial intelligence amyloid detection algorithm ↗ Precision Cardiac Amyloid · G Duffy, EK Oikonomou, N Easton, H Usuku, J Patel, Y Katsumata, ... Journal
2025 JACC: Advances Evaluating the performance and potential bias of predictive models for detection of transthyretin cardiac amyloidosis ↗ Independent head-to-head · J Hourmozdi, N Easton, S Benigeri, JD Thomas, A Narang, D Ouyang, ... Journal
2025 IEEE/CVF ICCV Workshops EchoNet-Quality: Denoising Echocardiograms via Deep Generative Modeling of Ultrasound Noise ↗ EchoNet-Quality · D Choi, M Vukadinovic, B He, C Binder, Y Sahashi, D Ouyang Conference
2025 European Heart Journal — Digital Health Automated evaluation for pericardial effusion and cardiac tamponade with echocardiographic artificial intelligence ↗ EchoNet-Pericardium · IM Chiu, Y Sahashi, M Vukadinovic, PP Cheng, S Cheng, D Ouyang Journal
2025 Current Cardiology Reports Current State of Artificial Intelligence in Assessing Cardiac Function Field review · V Yuan, K Lee, AP Ambrosy, D Ouyang, H Ieki Journal
2024 npj Digital Medicine Deep learning evaluation of echocardiograms to identify occult atrial fibrillation ↗ Occult atrial fibrillation · N Yuan, NR Stein, G Duffy, RK Sandhu, SS Chugh, PS Chen, ... Journal
2024 The Journal of Emergency Medicine AI-enabled assessment of cardiac function and video quality in emergency department point- of-care echocardiograms Point-of-care ultrasound · B He, D Dash, Y Duanmu, TX Tan, D Ouyang, J Zou Journal
2024 Scientific Reports Deep learning for transesophageal echocardiography view classification ↗ TEE view classification · KR Steffner, M Christensen, G Gill, M Bowdish, J Rhee, A Kumaresan, ... Journal
2024 Pacific Symposium on Biocomputing Leveraging 3D echocardiograms to evaluate AI model performance in predicting cardiac function on Out-of-Distribution data 3D model evaluation · G Duffy, K Christensen, D Ouyang Conference
2024 Open Heart Detection of cardiac amyloidosis using machine learning on routine echocardiographic measurements Amyloidosis from measurements · RSW Chang, I Chiu, P Tacon, M Abiragi, L Cao, G Hong, J Le, J Zou, ... Journal
2024 Nature Medicine Vision–language foundation model for echocardiogram interpretation ↗ EchoCLIP · M Christensen, M Vukadinovic, N Yuan, D Ouyang Journal
2024 Journal of the American College of Cardiology Random Forest Machine Learning to Detect Cardiac Amyloidosis Amyloidosis random forest · RSW Chang, PR Tacon, M Abiragi, L Cao, G Hong, J Le, P Ricchiuto, ... Abstract
2024 Journal of the American College of Cardiology DEEP LEARNING FOR HIGH THROUGHPUT AUTOMATED CALCULATION OF LVEF AND GLS Automated LVEF and GLS · G Hong, PR Tacon, J Le, L Cao, M Abiragi, RSW Chang, IM Chiu, ... Abstract
2024 JAMA Cardiology A multimodal video-based AI biomarker for aortic stenosis development and progression ↗ Aortic stenosis progression · EK Oikonomou, G Holste, N Yuan, A Coppi, RL McNamara, NA Haynes, ... Journal
2024 JACC: Cardiovascular Imaging Deep learning-derived myocardial strain ↗ Myocardial strain · AC Kwan, EW Chang, I Jain, J Theurer, X Tang, N Francisco, F Haddad, ... Journal
2024 JACC: Advances Navigating the gray zone: AI decision support to identify aortic stenosis severity ↗ Aortic stenosis severity · A Sarraju, D Ouyang Journal
2024 JACC: Advances Impact of case and control selection on training artificial intelligence screening of cardiac amyloidosis ↗ Cohort construction · A Vrudhula, L Stern, PC Cheng, P Ricchiuto, C Daluwatte, R Witteles, ... Journal
2024 European Heart Journal — Digital Health Clinical and genetic associations of asymmetric apical and septal left ventricular hypertrophy LVH genetic associations · V Yuan, M Vukadinovic, AC Kwan, F Rader, D Li, D Ouyang Journal
2024 Circulation External Validation of EchoNet-LVH, a Deep Learning Model for Cardiac Amyloidosis, for Association with Cardiomyopathy EchoNet-LVH external validation · PMM Castellote, G Duffy, W Zhou, S Cheng, J Chen, N Viney, S Tsimikas, ... Abstract
2024 Circulation High-throughput deep learning detection of mitral regurgitation ↗ EchoNet-MR · A Vrudhula, G Duffy, M Vukadinovic, D Liang, S Cheng, D Ouyang Journal
2024 Blood Evaluation of Functional Cardiac Measures and Response to Treatment Initiation in Patients with Systemic Light-Chain (AL) Amyloidosis: Results from a Single Site Retrospective … AL amyloidosis treatment response · J Thompson, J Catini, D Ouyang, IM Chiu, CC Quarta Abstract
2023 Pacific Symposium on Biocomputing Impact of Measurement Noise on Genetic Association Studies of Cardiac Function Measurement noise · M Vukadinovic, G Renjith, V Yuan, A Kwan, SC Cheng, D Li, SL Clarke, ... Conference
2023 Nature Blinded, randomized trial of sonographer versus AI cardiac function assessment ↗ EchoNet-RCT · B He, AC Kwan, JH Cho, N Yuan, C Pollick, T Shiota, J Ebinger, NA Bello, ... Journal
2023 Journal of the American Society of Echocardiography Prediction of coronary artery calcium using deep learning of echocardiograms ↗ Coronary artery calcium · N Yuan, AC Kwan, G Duffy, J Theurer, JH Chen, K Nieman, P Botting, ... Journal
2023 Journal of the American Society of Echocardiography Video-Based Deep Learning for Automated Assessment of Left Ventricular Ejection Fraction in Pediatric Patients ↗ EchoNet-Peds · CD Reddy, L Lopez, D Ouyang, JY Zou, B He Journal
2023 European Heart Journal Severe aortic stenosis detection by deep learning applied to echocardiography ↗ Severe aortic stenosis · G Holste, EK Oikonomou, BJ Mortazavi, A Coppi, KF Faridi, EJ Miller, ... Journal
2022 arXiv Deep learning discovery of demographic biomarkers in echocardiography ↗ Demographic biomarkers · G Duffy, SL Clarke, M Christensen, B He, N Yuan, S Cheng, D Ouyang Preprint
2022 JAMA Cardiology High-throughput precision phenotyping of left ventricular hypertrophy with cardiovascular deep learning ↗ EchoNet-LVH · G Duffy, PP Cheng, N Yuan, B He, AC Kwan, MJ Shun-Shin, ... Journal
2022 JACC: Cardiovascular Imaging Revival and revision of right ventricular assessment by machine learning Right ventricular assessment · D Ouyang, S Cheng Journal
2022 European Heart Journal — Digital Health Multimodal deep learning enhances diagnostic precision in left ventricular hypertrophy ↗ LVH-fusion · JT Soto, J Weston Hughes, PA Sanchez, M Perez, D Ouyang, EA Ashley Journal
2021 Pacific Symposium on Biocomputing Interpretable deep learning prediction of 3d assessment of cardiac function 3D cardiac function · G Duffy, I Jain, B He, D Ouyang Conference
2021 JACC: Cardiovascular Imaging Characterizing mitral regurgitation with precision phenotyping and unsupervised learning Mitral regurgitation phenotyping · D Ouyang, JD Thomas Journal
2021 JACC: Cardiovascular Imaging Systematic quantification of sources of variation in ejection fraction calculation using deep learning EF measurement variability · N Yuan, I Jain, N Rattehalli, B He, C Pollick, D Liang, P Heidenreich, ... Journal
2021 EBioMedicine Deep learning evaluation of biomarkers from echocardiogram videos ↗ EchoNet-Labs · JW Hughes, N Yuan, B He, J Ouyang, J Ebinger, P Botting, J Lee, ... Journal
2021 Circulation Video-based deep learning model for automated assessment of ejection fraction in pediatric patients EchoNet-Peds · B He, D Ouyang, L Lopez, J Zou, CD Reddy Abstract
2020 npj Digital Medicine Deep learning interpretation of echocardiograms ↗ Multi-task echo interpretation · A Ghorbani, D Ouyang, A Abid, B He, JH Chen, RA Harrington, DH Liang, ... Journal
2020 Nature Video-based AI for beat-to-beat assessment of cardiac function ↗ EchoNet-Dynamic · D Ouyang, B He, A Ghorbani, N Yuan, J Ebinger, CP Langlotz, ... Journal
2020 Journal of the American College of Cardiology A deep learning algorithm accurately detects pericardial effusion on echocardiography Pericardial effusion · A Nayak, D Ouyang, EA Ashley Abstract
2019 NeurIPS ML4H Workshop Echonet-dynamic: a large new cardiac motion video data resource for medical machine learning EchoNet-Dynamic (dataset) · D Ouyang, B He, A Ghorbani, MP Lungren, EA Ashley, DH Liang, JY Zou Conference

Open science

The model family behind these products is published and reproducible, and the underlying annotated datasets were released publicly — including 23,212 annotated echocardiogram videos from the JAMA Cardiology 2022 study.

That has a consequence most vendors avoid: independent groups can evaluate these models without asking us, and they have. A Northwestern-led team ran a head-to-head against another FDA-cleared device and published a fairness audit alongside it. We host that study, in full, including the metrics where we score lower — see the amyloid evidence page.

A note on names. EchoNet is the research lab, not a single model. Each paper carries its own name — EchoNet-Dynamic for ejection fraction, EchoNet-LVH for hypertrophy and amyloidosis, EchoNet-Labs for laboratory values, EchoNet-MR for mitral regurgitation, EchoNet-Measurements, EchoNet-Liver, EchoNet-Quality, alongside EchoCLIP and EchoPrime. The FDA clearances use product names for the same models: EchoNet-Dynamic is InVision Precision LVEF, and EchoNet-LVH is InVision Precision Cardiac Amyloid.

How to read an AUC

The same model reports different AUCs across these papers. That is not instability — it is cohort construction, and we published the paper that demonstrates it.

In JACC: Advances (2024), model AUCs ranged from 0.660 to 0.898 on matched held-out test sets and 0.467 to 0.898 in a general patient population — varying by nothing except how cases and controls were selected. An AUC is a property of a model and a cohort. Comparing one study's AUC against another's is not a valid comparison.

Questions about the research?

Our clinical team includes authors on most of these papers. They will take any of it as deep as you want to go.

We reply within one business day.
Publications render from _data/claims.json · Last reviewed August 2026.