Heart failure & cardiomyopathiesDiagnostic or prognostic study
Cardiac amyloidosis detection from a single echocardiographic video clip: a novel artificial intelligence-based screening tool
- Diagnostic accuracy study
- multisite
- external validation across 18 sites
Authors' conclusion
This AI screening model-using only an apical four-chamber view-effectively differentiated CA from other causes of increased left ventricular wall thickness.
Study context
- Population
- Cardiac amyloidosis (CA) vs phenotypic controls; external validation 597 CA cases and 2122 controls597 CA, 2122 controls
- Comparison
- Artificial intelligence (AI) echocardiography model vs transthyretin CA score and increased wall thickness score
- Outcome
- Classification accuracy for CA detection on the external validation dataset
Performance
As reported in the abstract
CA detection, entire external validation dataset
| Test or model | AUC | Sensitivity | Specificity |
|---|---|---|---|
| AI model | 0.93 | 85% | 93% |
Further findings
AUROC
| Outcome | AI model | Transthyretin CA score | Increased wall thickness score |
|---|---|---|---|
| AUROC vs clinical scores (older HFpEF patients with increased wall thickness, subset) | 0.93 | 0.73 | 0.80 |
Sensitivity by CA subtype (external validation)
- Sensitivity, light-chain
- 84%
- Sensitivity, wild-type transthyretin
- 85%
- Sensitivity, hereditary transthyretin
- 86%
Performance in patients referred for technetium pyrophosphate scintigraphy (subgroup)
- AUROC
- 0.86
- Sensitivity
- 77%
- Specificity
- 86%
Performance in patients matched for age, sex and wall thickness (subgroup)
- AUROC
- 0.92
- Sensitivity
- 84%
- Specificity
- 91%
- matched patients
Original publication
Cardiac amyloidosis detection from a single echocardiographic video clip: a novel artificial intelligence-based screening tool. Eur Heart J. Published 9 July 2025. doi:10.1093/eurheartj/ehaf387
Summarised from the published abstract; every number is checked against it