Diagnostic or prognostic study
Machine Learning to Optimize the Echocardiographic Follow-Up of Aortic Stenosis
- Prognostic model development and validation study
- internal validation
- external validation cohort
Authors' conclusion
Machine learning provides real-time, automated, personalized timing of next echocardiographic follow-up examination for patients with mild-to-moderate aortic stenosis. Compared with European and American guidelines, the model reduces the number of patient examinations.
Study context
- Population
- Patients with mild-to-moderate AS
- Participants
- 1,638 vs 1,533
- Comparison
- Internal validation vs External application
- Internal validationn = 1,638
- External applicationn = 1,533
- Outcome
- Prediction of development of severe AS
- Follow-up
- 1, 2 or 3 years
Performance
As reported in the abstract
Discrimination of severe vs nonsevere AS development (AUC) · 1-, 2-, 3-year interval
| Test or model | AUC 1 year | AUC 2 years | AUC 3 years |
|---|---|---|---|
| Internal validation | 0.90 | 0.92 | 0.92 |
| External application | 0.85 | 0.85 | 0.85 |
Further findings
Unnecessary echocardiographic examinations saved per year (simulated, external cohort)
- Savings vs European guideline recommendations
- 49%
- Savings vs American guideline recommendations
- 13%
- simulated application
Original publication
Machine Learning to Optimize the Echocardiographic Follow-Up of Aortic Stenosis. JACC Cardiovasc Imaging. Published 8 February 2023. doi:10.1016/j.jcmg.2022.12.008
Summarised from the published abstract; every number is checked against it