Valvular & structural heart disease
Integrating Echocardiography Parameters With Explainable Artificial Intelligence for Data-Driven Clustering of Primary Mitral Regurgitation Phenotypes
Authors' conclusion
Novel data-driven phenogrouping and explainable AI aided in improved integration of echocardiographic data to identify patients with primary MR and improved event-free survival after mitral valve repair/replacement surgery.
Study context
- Population
- Primary mitral regurgitation (MR) subjects from France (development cohort) and Canada (validation cohort)France 243
- Participants
- 400 subjects
- Exposure
- Surgery vs no surgery within data-driven high- and low-severity phenogroups
- Outcome
- All-cause mortality, with time-to-mitral valve surgery as time-dependent covariate
Methodological detailsunsupervised and supervised machine learning, explainable artificial intelligence (AI), development and validation cohorts, preliminary study, Median follow-up 3.2 years (IQR: 1.3-5.3) France; 6.8 (IQR: 4.0-8.5 years) Canada
- Unsupervised and supervised machine learning
- Explainable AI
- Development and validation cohorts
- Preliminary study
- Median follow-up 3.2 years (IQR: 1.3-5.3) France; 6.8 (IQR: 4.0-8.5 years) Canada
Main finding
As reported in the abstract
Event-free survival, surgical vs nonsurgical, by phenogroup
- French high-severity (HS): improved event-free survival with surgery vs no surgery
- P = 0.047
- Canadian HS: improved event-free survival with surgery vs no surgery
- P = 0.020
- French LS: no similar benefit of surgery
- P = 0.70
- Canadian LS: no similar benefit of surgery
- P = 0.50
Observational data: associations, not effects.
Further findings
Incremental prognostic value of phenogrouping in conventionally severe or moderate-severe MR
- Harrell C statistic improvement
- P = 0.480
- Categorical net reclassification improvement
- P = 0.002
Original publication
Integrating Echocardiography Parameters With Explainable Artificial Intelligence for Data-Driven Clustering of Primary Mitral Regurgitation Phenotypes. JACC Cardiovasc Imaging. Published 10 May 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →