Heart failure & cardiomyopathies
Machine Learning in Hypertrophic Cardiomyopathy: Nonlinear Model From Clinical and Cardiac Magnetic Resonance (CMR) Features Predicting Cardiovascular Events
Performance · Prediction of MACEResult — go to the resultImprovement in AUC 22.7%Authors' conclusion
ML-empowered risk stratification using CMR and clinical features enabled accurate MACE prediction beyond the classic HCM Risk-SCD model.
Study context
- Population
- Patients with hypertrophic cardiomyopathy (HCM) admitted 2010–2017 at 4 medical centres (China); internal discovery cohort 533, external test cohort 225
- Participants
- 758 patients4 centres
- Comparison
- Machine learning (ML) model vs HCM Risk-SCD model
- Outcome
- MACE:
- Arrhythmic events
- Sudden cardiac death (SCD)
- HF
- AF-related stroke
- Follow-up
- Median 109.0 months (Q1-Q3: 73.0-118.8 months)
Methodological detailsmulticentre, internal discovery cohort with cross-validation, external test cohort
- Multicentre
- Internal discovery cohort with cross-validation
- External test cohort
Performance
Prediction of MACE
Improvement in AUC
22.7%
P < 0.001
By group
ML model
- AUC (internal)
- 0.830
- AUC (external)
- 0.812
ML model
AUC (internal)0.830
AUC (external)0.812
As reported in the abstract
Further findings
MACE incidence · median follow-up 109.0 months
- MACE occurred
- 191 (25%)
Nonlinear association of LGE, global radial strain (GRS) and global circumferential strain (GCS) with MACE (cubic spline)
- Elevated risk above second-tertile threshold, LGE
- 11.6%
- Elevated risk above second-tertile threshold, GRS
- 25.8%
- Elevated risk above second-tertile threshold, GCS
- −17.3%
Original publication
Machine Learning in Hypertrophic Cardiomyopathy: Nonlinear Model From Clinical and CMR Features Predicting Cardiovascular Events. JACC Cardiovasc Imaging. Published 10 July 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →