Machine Learning for Prediction of Adverse Cardiovascular Events in Adults With Repaired Tetralogy of Fallot Using Clinical and Cardiovascular Magnetic Resonance Imaging Variables
Authors' conclusion
In this single-center study, a machine learning-based prediction model comprised of readily available clinical and cardiovascular magnetic resonance imaging variables performed well in an independent validation cohort.
Study context
- Population
- Adults with repaired tetralogy of Fallot with major adverse cardiovascular events (MACE) or followed ≥5 years312 development, 492 validation
- Participants
- 804 adults
- Comparison
- Random forest ML model vs conventional Cox multivariable model
- Outcome
- Prediction of MACE:
- Mortality
- Resuscitated sudden death
- Sustained VT
- HF
- Follow-up
- 5 years
Methodological detailssingle-centre, random forest machine learning, independent validation cohort
- Single-centre
- Random forest machine learning
- Independent validation cohort
Performance
As reported in the abstract
MACE prediction in validation dataset
| Test or model | AUC |
|---|---|
| ML model | 0.82 [0.74-0.89] |
| Cox model | 0.63 [0.51-0.75] |
AUC
ML model0.82 [0.74-0.89]
Cox model0.63 [0.51-0.75]
- ML model vs Cox model: P = 0.003 — superior performance of ML model
Further findings
AUC for MACE with input restricted to the 10 strongest features (validation dataset) · 5 years
- Model performance did not change significantly with input restricted to the 10 strongest features
- 0.81 [0.72-0.89]; P=0.232
AUC for MACE after removing exercise parameters (validation dataset) · 5 years
- Removing exercise parameters resulted in inferior model performance
- 0.75 [0.65-0.84]; P=0.002
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Machine Learning for Prediction of Adverse Cardiovascular Events in Adults With Repaired Tetralogy of Fallot Using Clinical and Cardiovascular Magnetic Resonance Imaging Variables. Circ Cardiovasc Imaging. Published 20 June 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →