Heart failure & cardiomyopathies
Machine Learning of Echocardiographic Left Ventricular Myocardial Radiomics for Identification of Transthyretin Amyloid Cardiomyopathy
- retrospective and prospective multicenter study
- development and validation of an echocardiography-based machine learning model
- independent internal validation set
- external validation set
+3 more
- development and validation of an echocardiography-based machine learning model
- independent internal validation set
- external validation set
Authors' conclusion
We developed an echocardiographic radiomics-based tool to identify ATTR-CM with excellent discrimination and very high NPV in a cohort that approximates clinical experience. These preliminary findings require validation in larger, more diverse cohorts.
Study context
- Population
- Patients with ATTR-CM and comparators (nonamyloid heart failure (HF), immunoglobulin light-chain-amyloid, hypertrophic cardiomyopathy) undergoing echocardiographyechocardiographic scans analyzed, training set (150 ATTR-CM, 150 nonamyloid HF), internal validation set (50 ATTR-CM, 399 nonamyloid HF), external validation set (10 ATTR-CM, 10 immunoglobulin light-chain-amyloid, 10 hypertrophic cardiomyopathy, 75 nonamyloid HF)
- Participants
- 854 patients
- Exposure
- Echocardiography-based machine learning model using myocardial radiomics to identify ATTR-CM
Main finding
As reported in the abstract
Model performance, internal validation
- sensitivity
- 86% (95% CI: 73.8%-93%)
- specificity
- 92% (95% CI: 88.9%-94.3%)
- positive predictive value
- 57.3% (95% CI: 46.1%-67.9%)
- NPV
- 98.1% (95% CI: 96.2%-99.1%)
- receiver operating characteristic-area under the curve
- 0.938 (95% CI: 0.891-0.985)
Observational data: associations, not effects.
Further findings
| Outcome | ATTR-CM | nonamyloid HF | AL-amyloid | hypertrophic cardiomyopathy | |
|---|---|---|---|---|---|
| Correct classification as non-ATTR (subgroup analysis) | Not reported in the abstract | 77.3% | 70% | 100% | ATTR-CM: not reported in the abstract |
Model performance, external validation
- sensitivity
- 100%
- specificity
- 78.9%
- positive predictive value
- 33.3%
- NPV
- 100%
- receiver operating characteristic-area under the curve
- 0.917
Original publication
Machine Learning of Echocardiographic Left Ventricular Myocardial Radiomics for Identification of Transthyretin Amyloid Cardiomyopathy. JACC Adv. Published 13 October 2026. doi:10.
Summarised from the published abstract; every number is checked against it