Valvular & structural heart disease
Data-driven mortality risk prediction of severe degenerative mitral regurgitation patients undergoing mitral valve surgery
Authors' conclusion
A data-driven machine learning model provided accurate post-MVR mortality prediction in severe degenerative MR patients.
Study context
- Population
- Consecutive patients with severe degenerative mitral regurgitation (MR) undergoing mitral valve replacement/repair
- Participants
- 1,521 patients
- Comparison
- Mitral valve replacement/repair (MVR)-score vs modified MIDA score
- Outcome
- All-cause mortality after MVR
- Follow-up
- 3 years
Methodological detailsmachine learning, random survival forest, 70% training/30% test sets
- Machine learning
- Random survival forest
- 70% training/30% test sets
Performance
As reported in the abstract
Prediction of 3-year post-MVR mortality
| Test or model | C-index (test set) |
|---|---|
| Random survival forest (RSF) model | 0.880, 95% CI 0.834-0.925 |
Random survival forest (RSF) model
C-index (test set)0.880, 95% CI 0.834-0.925
- Deaths at 3 years, entire cohort
- 90 (5.9%)
- Deaths in training and test sets
- 59 and 31
- Mortality risk increased strongly with left atrial dimension
- >55 mm
- Mortality risk increased strongly with left ventricular (LV) end-systolic diameter
- >45 mm
Further findings
| Outcome | MVR-score | P value |
|---|---|---|
| Predictability, MVR-score vs modified MIDA score · 3 years | ||
| C-index | 0.803 | P = 0.034 |
- Predictability, MVR-score vs modified MIDA score: Modified MIDA score C-index 0.750
Original publication
Data-driven mortality risk prediction of severe degenerative mitral regurgitation patients undergoing mitral valve surgery. Eur Heart J Cardiovasc Imaging. Published 28 April 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →