The deep-learning AI model Us2.ca detected cardiac amyloidosis with 87.5% accuracy and AUC 0.92, outperforming the AI-derived multiparametric echocardiographic score accuracy of 79.5% and AUC 0.87 in external US validation.
Observational (n=5,776)
Yes
Does an AI-based deep-learning model accurately diagnose cardiac amyloidosis on echocardiography compared to an AI-derived multiparametric score in patients with suspected hypertrophy?
A fully automated deep-learning model applied to echocardiography videos can accurately identify cardiac amyloidosis across diverse global cohorts, outperforming AI-derived multiparametric scores.
Effect estimate: AUC 0.92 for Us2.ca model vs AUC 0.87 for multiparametric score in US cohort and AUC 0.93 vs 0.85 in Japan cohort (95% CI 95% CI 0.90–0.94 for Us2.ca (US), 0.84–0.90 for score (US), 0.91–0.96 for Us2.ca (Japan), 0.81–0.90 for score (Japan))
Absolute Event Rate: 87.5% vs 79.5%
p-value: p=<0.001
BACKGROUND: Diagnosing cardiac amyloidosis (CA) on echocardiography can be challenging due to the imaging overlap between CA and more prevalent causes of a hypertrophic phenotype. This study sought to (1) evaluate the performance of artificial-intelligence (AI) derived measurements incorporated into the established multiparametric echocardiographic scoring system to detect CA; (2) develop and validate an AI-based deep-learning model for video-based detection of CA on echocardiography. METHODS: The study population comprised 5776 patients (CA, 2756; controls, 3020). The training data set included patients from the UK National Amyloidosis Center and Taiwan MacKay Memorial Hospital (CA, 2241; controls, 2130). External test data sets were obtained from the US Duke University Health System (CA, 334; LVH controls, 668) and Japan National Cerebral and Cardiovascular Center (CA, 181; LVH controls, 222). RESULTS: The multiparametric echocardiographic score computed using AI-derived measurements achieved an accuracy of 79.5% (sensitivity, 75.4%; specificity, 81.5%) in the United States cohort and 79.7% (sensitivity, 81.6%; specificity, 78.1%) in the Japan cohort. The deep-learning model demonstrated accuracies of 96.2% (sensitivity, 96.8%; specificity, 95.7%) and 95.8% (sensitivity, 97.3%; specificity, 94.3%) in the internal validation and internal test sets, respectively. External validation of the deep-learning model showed accuracies of 87.5% (sensitivity, 86.6%; specificity, 87.9%) in the United States and 88.4% (sensitivity, 92.3%; specificity, 85.3%) in the Japanese cohort. Subgroup analysis demonstrated that the deep-learning model showed robust discrimination of CA from other hypertrophic phenocopies: CA versus hypertension (area under the curve AUC, 0.92 95% CI, 0.91–0.94), CA versus hypertrophic cardiomyopathy (AUC, 0.91 95% CI, 0.87–0.94), CA versus aortic stenosis (AUC, 0.93 95% CI, 0.90–0.95), CA versus chronic kidney disease (AUC, 0.93 95% CI, 0.91–0.95). The deep-learning model was able to classify a greater proportion of patients compared with the AI-derived multiparametric echocardiographic score and achieved superior diagnostic accuracy (AUC, 0.93 95% CI, 0.91–0.95 versus AUC, 0.88 95% CI, 0.85–0.90; P <0.001). CONCLUSIONS: Both the multiparametric echocardiographic score computed from AI-derived measurements and the fully automated deep-learning model can accurately identify patients with CA in globally diverse cohorts, with the deep-learning model providing superior performance.
Ioannou et al. (2026) conducted an observational in Patients with suspected cardiac amyloidosis and controls with left ventricular hypertrophy across multiple international centers (n=5,776). Deep-learning AI-based video model (Us2.ca) for detection of cardiac amyloidosis on echocardiography vs. AI-derived multiparametric echocardiographic score; standard clinical assessment was evaluated on Diagnostic accuracy of cardiac amyloidosis detection on echocardiography (AUC 0.92 for Us2.ca model vs AUC 0.87 for multiparametric score in US cohort and AUC 0.93 vs 0.85 in Japan cohort, 95% CI 95% CI 0.90–0.94 for Us2.ca (US), 0.84–0.90 for score (US), 0.91–0.96 for Us2.ca (Japan), 0.81–0.90 for score (Japan), p=<0.001). The deep-learning AI model Us2.ca detected cardiac amyloidosis with 87.5% accuracy and AUC 0.92, outperforming the AI-derived multiparametric echocardiographic score accuracy of 79.5% and AUC 0.87 in external US validation.