Valvular & structural heart disease
Deep learning for echocardiographic assessment and risk stratification of aortic, mitral, and tricuspid regurgitation: the DELINEATE-regurgitation study
Authors' conclusion
An AI system can accurately classify AR, MR, and TR and predict MR progression beyond currently known risk factors.
Study context
- Population
- Transthoracic echocardiograms with colour Doppler videos assessed for aortic, mitral and tricuspid regurgitation
- Participants
- 71,660 TTEs
- Comparison
- DELINEATE-Regurgitation artificial intelligence (AI) vs cardiologist interpretation
- Outcome
- Agreement of AI regurgitation severity classification with cardiologists (weighted kappa)
Methodological detailstwo sites, internal and external test sets, deep learning
- Two sites
- Internal and external test sets
- Deep learning
Performance
As reported in the abstract
Weighted kappa, multiview approach
| Test or model | Aortic (AR) | Mitral (MR) | Tricuspid regurgitation (TR) |
|---|---|---|---|
| Internal test set | 0.81 | 0.76 | 0.73 |
| External test set | 0.76 | 0.72 | 0.64 |
Aortic (AR)
Internal test set0.81
External test set0.76
Mitral (MR)
Internal test set0.76
External test set0.72
Tricuspid regurgitation (TR)
Internal test set0.73
External test set0.64
Further findings
MR progression to moderate-severe or severe MR (DELINEATE-MR-Progression AI score), adjusted for clinical factors
| Comparison | HR (95% CI) | |
|---|---|---|
| AI score · internal test set | 4.1 (2.5–6.6) | |
| 0.10.31310Hazard ratio (log scale) |
Original publication
Deep learning for echocardiographic assessment and risk stratification of aortic, mitral, and tricuspid regurgitation: the DELINEATE-regurgitation study. Eur Heart J. Published 29 March 2025. doi:10.
Summarised from the published abstract; every number is checked against it