Heart failure & cardiomyopathies
International Validation of Echocardiographic Artificial Intelligence Amyloid Detection Algorithm
Authors' conclusion
EchoNet-LVH can assist with earlier and accurate diagnosis of CA.
Study context
- Population
- Echocardiogram studies of cardiac amyloidosis (CA) patients and controls across multiple international sites574 CA, 979 controls
- Outcome
- Discrimination of CA vs controls (AUC; sensitivity, specificity, PPV at prespecified threshold)
Methodological detailsmultisite, retrospective case-control, prespecified threshold
- Multisite
- Retrospective case-control
- Prespecified threshold
Performance
As reported in the abstract
CA detection
| Test or model | AUC | Sensitivity | Specificity | PPV | NPV |
|---|---|---|---|---|---|
| EchoNet-LVH | 0.896 (95% CI: 0.875-0.916) | 0.644 (95% CI: 0.601-0.685) | 0.988 (95% CI: 0.978-0.994) | 0.968 (95% CI: 0.944-0.984) | 0.828 (95% CI: 0.804-0.850) |
EchoNet-LVH
AUC0.896 (95% CI: 0.875-0.916)
Sensitivity0.644 (95% CI: 0.601-0.685)
Specificity0.988 (95% CI: 0.978-0.994)
PPV0.968 (95% CI: 0.944-0.984)
NPV0.828 (95% CI: 0.804-0.850)
Further findings
Heterogeneity by site, race, sex, age, BMI, CA subtype, ultrasound manufacturer
- No evidence of heterogeneity in performance
Original publication
International Validation of Echocardiographic Artificial Intelligence Amyloid Detection Algorithm. JACC Adv. Published 17 September 2025. doi:10.
Summarised from the published abstract; every number is checked against it