Valvular & structural heart disease
Artificial Intelligence–Enabled Acquisition and Interpretation for Screening Aortic Stenosis
Authors' conclusion
A deep learning algorithm accurately detected moderate or greater AS across validation cohorts in this study. In prospective evaluation, novice operators were able to acquire AI-guided FoCUS examinations that enabled accurate detection of moderate or greater AS.
Study context
- Population
- Patients at Mayo Clinic sites in the Midwest, Arizona and Florida assessed for moderate or greater aortic stenosis (AS)6,753 for model development
- Participants
- 844 / 1,912 / 602 / 1,302
- Comparison
- Internal test vs Validation cohorts vs Experienced sonographers vs Novice operators
- Internal testn = 844
- Validation cohortsn = 1,912
- Experienced sonographersn = 602
- Novice operatorsn = 1,302
- Outcome
- Detection of moderate or greater AS (AUROC, sensitivity, specificity, PPV)
Methodological detailsretrospective development and validation, prospective evaluation, multicentre
- Retrospective development and validation
- Prospective evaluation
- Multicentre
Performance
As reported in the abstract
Detection of moderate or greater AS
| Test or model | AUROC | AUROC, Arizona | AUROC, Florida | Sensitivity | Specificity |
|---|---|---|---|---|---|
| Internal test | 0.99 (95% CI, 0.98-1.00) | Not reported in the abstract | Not reported in the abstract | Not reported in the abstract | Not reported in the abstract |
| Validation cohorts | Not reported in the abstract | 0.99 (95% CI, 0.97-1.00) | 0.99 (95% CI, 0.96-1.00) | Not reported in the abstract | Not reported in the abstract |
| Experienced sonographers | Not reported in the abstract | Not reported in the abstract | Not reported in the abstract | 95% (95% CI, 82-99) | 97% (95% CI, 95-98) |
AUROC
Internal test0.99 (95% CI, 0.98-1.00)
Validation cohortsNot reported in the abstract
Experienced sonographersNot reported in the abstract
AUROC, Arizona
Internal testNot reported in the abstract
Validation cohorts0.99 (95% CI, 0.97-1.00)
Experienced sonographersNot reported in the abstract
AUROC, Florida
Internal testNot reported in the abstract
Validation cohorts0.99 (95% CI, 0.96-1.00)
Experienced sonographersNot reported in the abstract
Sensitivity
Internal testNot reported in the abstract
Validation cohortsNot reported in the abstract
Experienced sonographers95% (95% CI, 82-99)
Specificity
Internal testNot reported in the abstract
Validation cohortsNot reported in the abstract
Experienced sonographers97% (95% CI, 95-98)
- Prospective novice-operator cohort: 1258 of 1302 examinations suitable for automated analysis
- 96.6%
- Prospective novice-operator cohort: sensitivity
- 93% (95% CI, 82-99)
- Prospective novice-operator cohort: specificity
- 96% (95% CI, 95-97)
Further findings
Expert review of artificial intelligence (AI)-positive and uninterpretable examinations (novice-operator cohort)
- Sensitivity with expert review
- 85.4%
- PPV increased from 49.4% to 91.1%
Original publication
Artificial Intelligence–Enabled Acquisition and Interpretation for Screening Aortic Stenosis. JAMA Cardiol. Published 28 August 2026. doi:10.
Summarised from the published abstract; every number is checked against it