Diagnostic or prognostic study
Vision–language foundation model for echocardiogram interpretation
- Model development and validation study
- external validation
Authors' conclusion
These capabilities represent a substantial step toward understanding and applying foundation models in cardiovascular imaging for preliminary interpretation of echocardiographic findings.
Study context
- Population
- Cardiac ultrasound videos with expert cardiologist interpretations across a wide range of patients and imaging indications1,032,975 training videos
- Comparison
- EchoCLIP vs EchoCLIP-R
Performance
As reported in the abstract
Performance of EchoCLIP on cardiac image interpretation benchmarks (LVEF prediction; detection of implanted intracardiac devices)
| Test or model | LVEF prediction, mean absolute error (external validation) | Pacemaker AUC | Percutaneous mitral valve repair AUC | Artificial aortic valve AUC |
|---|---|---|---|---|
| EchoCLIP | 7.1% | 0.84 | 0.92 | 0.97 |
Original publication
Vision–language foundation model for echocardiogram interpretation. Nat Med. Published 30 April 2024. doi:10.1038/s41591-024-02959-y
Summarised from the published abstract; every number is checked against it