Valvular & structural heart diseaseDiagnostic or prognostic study
An Automated Machine Learning–Based Quantitative Multiparametric Approach for Mitral Regurgitation Severity Grading
- Diagnostic accuracy study
- development and external validation
- fully automated machine learning (ML) echocardiography workflow
- echocardiography core laboratory ground truth
Authors' conclusion
An automated multiparametric ML model for grading MR severity is feasible, fast, highly accurate, and predicts 1-year mortality.
Study context
- Population
- Patients with echocardiograms from 2 observational cohorts (training) and 2 additional independent studies (validation), graded for mitral regurgitation (MR) severity
- Comparison
- Automated ML MR severity grading vs core laboratory multiparametric assessment
- Outcome
- Accuracy of automated ML grading of MR severity
Performance
As reported in the abstract
MR severity grading vs core laboratory ground truth
Accuracy, MR severity (none to severe)
0.80
Accuracy, significant (moderate or severe) vs nonsignificant MR
0.97
Sensitivity, significant MR
0.96
Specificity, significant MR
0.98
Further findings
| Outcome | adjusted HR (95% CI) | P value | |
|---|---|---|---|
| 1-year mortality by ML-graded MR severity | 5.20 (1.24–21.9) | P = 0.025 | |
| 0.050.21520Adjusted hazard ratio (log scale) |
Feasibility and analysis time
- Image analysis feasible
- 99.3%
- Time per case (seconds)
- 80 ± 5
Eccentric vs central MR
- Model performed similarly in eccentric and central MR
Original publication
An Automated Machine Learning–Based Quantitative Multiparametric Approach for Mitral Regurgitation Severity Grading. JACC Cardiovasc Imaging. Published 14 August 2024. doi:10.1016/j.jcmg.2024.06.011
Summarised from the published abstract; every number is checked against it