Valvular & structural heart disease
Multimodal Machine Learning-Based Technical Failure Prediction in Patients Undergoing Transcatheter Aortic Valve Replacement
Authors' conclusion
Machine learning-based models that integrate multimodal data can effectively predict VARC-3 technical failure in TAVR
Study context
- Population
- Consecutive patients undergoing transcatheter aortic valve replacement (TAVR) in a prospective TAVR registry
- Participants
- 2,937 patients
- Outcome
- VARC-3 technical failure (cardiac and vascular)
Methodological detailsprospective registry, machine learning, holdout test set
- Prospective registry
- Machine learning
- Holdout test set
Performance
As reported in the abstract
VARC-3 technical failure prediction (best-performing model, holdout test set)
| Test or model | AUC | NPV |
|---|---|---|
| Cardiac technical failure model | 0.769 | 0.995 |
| Vascular technical failure model | 0.788 | 0.976 |
AUC
Cardiac technical failure model0.769
Vascular technical failure model0.788
NPV
Cardiac technical failure model0.995
Vascular technical failure model0.976
Further findings
Rate of technical failure
- Cardiac technical failure
- 2.4%
- Vascular technical failure
- 7.0%
Key determinants of technical failure (interpretability analysis)
- Atherosclerotic comorbidities, CT-based aortic root and iliofemoral anatomies, antithrombotic management and procedural features consistently identified across all models
Original publication
Multimodal Machine Learning-Based Technical Failure Prediction in Patients Undergoing Transcatheter Aortic Valve Replacement. JACC Adv. Published 18 September 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →