Differentiating cardiac amyloidosis (CA) from its phenotypic mimics remains a clinical challenge. A new study published in the European Heart Journal details the development and validation of a novel artificial intelligence (AI) model that can accurately screen for CA using only a single apical four-chamber echocardiogram video clip.¹
This retrospective, multisite, multiethnic study first trained a convolutional neural network (CNN) using a dataset of 2,612 patients (52% with CA). The model was trained to distinguish CA from other causes of increased left ventricular wall thickness, such as hypertrophic cardiomyopathy and hypertensive heart disease.
The AI model’s performance was then assessed in a separate, international external validation cohort comprising 2,719 patients from 18 global sites, including 597 confirmed CA cases and 2,122 controls. Subgroup analyses were conducted on patients referred for technetium pyrophosphate (Tc-PYP) scintigraphy and in a cohort matched for age, sex, and wall thickness. The model’s accuracy was also compared with the transthyretin CA score (TCAS) and the increased wall thickness (IWT) score.
After excluding uncertain predictions (13.4% of cases), the AI model demonstrated excellent diagnostic performance in the external validation cohort. It achieved an area under the receiver operating characteristic curve (AUROC) of 0.93, with a sensitivity of 85% and a specificity of 93%.
The model’s accuracy was consistent across different CA subtypes, including light-chain (AL-CA), wild-type transthyretin (ATTRwt-CA), and hereditary transthyretin (ATTRv-CA) amyloidosis. Performance was maintained in the challenging subgroup of patients referred for Tc-PYP imaging (AUROC 0.86) and in the phenotypically similar matched cohort (AUROC 0.92).
Furthermore, the AI model significantly outperformed existing clinical risk scores in a subset of older patients with heart failure with preserved ejection fraction (HFpEF) and increased wall thickness, achieving an AUROC of 0.92 compared with 0.74 for the TCAS and 0.80 for the IWT score.
These findings suggest that a fully automated AI tool applied to a routine echocardiographic view can effectively screen for CA with high accuracy. This approach has the potential to improve the efficiency of the diagnostic pathway for CA, helping to identify patients who may benefit from further specific testing and facilitating earlier access to life-prolonging therapies.
The study authors noted that prospective trials are needed to confirm the model's clinical utility and to determine how it can be best integrated into existing diagnostic workflows.
This study was funded by Ultromics.
References
1. Slivnick JA, Hawkes W, Oliveira J, et al. Cardiac amyloidosis detection from a single echocardiographic video clip: a novel artificial intelligence-based screening tool. Eur Heart J 2025;46:4090–4101. https://doi.org/10.1093/eurheartj/ehaf387
Disclaimer: The information presented in this article is for educational purposes only. Any quotes included reflect the opinions of the individual quoted, and do not necessarily reflect the views of the publisher. The publisher does not guarantee the accuracy or completeness of the content and accepts no responsibility for any errors, or any consequences arising from its use.