A multimodal AI model using chest CT, echocardiography, and ECG achieved an AUROC of 0.85 (95% CI 0.74-0.93) for detecting ATTR-CM in TAVR patients, outperforming single-modality approaches.
Observational (n=816)
Yes
Does a multimodal AI model using chest CT, echocardiography, and electrocardiography improve the detection of ATTR-CM in patients undergoing TAVR?
816 patients who underwent technetium-99m–pyrophosphate scintigraphy (PYP) at two academic medical centers, median age 79.0 years, 61.2% male.
Multimodal artificial intelligence (AI) model using chest computed tomography (CT), echocardiography, and electrocardiography
Single-modality approaches
Diagnostic performance for detecting ATTR-CM measured by the area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, and predictive valuessurrogate
A multimodal AI approach using routine chest CT, echocardiography, and ECG can effectively screen for ATTR-CM in TAVR patients, potentially facilitating earlier diagnosis.
Effect estimate: AUROC 0.85 (95% CI 0.74-0.93)
Abstract Background and Aims Early detection is important given the availability of new disease-modifying therapies and the high prevalence of transthyretin amyloid cardiomyopathy (ATTR-CM) among patients with aortic stenosis (AS) undergoing transcatheter aortic valve replacement (TAVR). We developed a multimodal artificial intelligence (AI) model for early detection of ATTR-CM using chest computed tomography (CT), echocardiography, and electrocardiography. This approach may provide a scalable strategy for preclinical monitoring. Methods This retrospective study included patients who underwent technetium-99m–pyrophosphate scintigraphy (PYP) at two academic medical centers: Columbia University Irving Medical Center and Weill Cornell Medicine. ATTR-CM status was determined using a composite reference standard incorporating PYP scan interpretation, laboratory tests, and endomyocardial biopsy results when available. The diagnostic performance of the model was measured by the area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, and predictive values at various thresholds. Results Among 816 patients (median age 79.0 years, 61.2% male), 127 (15.6%) had confirmed ATTR-CM. Patients with ATTR-CM were older, more often male, and had characteristic echocardiographic features, including increased wall thickness and reduced ejection fraction. In the independent TAVR test cohort, the multimodal AI model achieved an AUROC of 0.85 (95% CI: 0.74-0.93), significantly outperforming single-modality approaches in our data. At the optimal threshold, the model demonstrated 73.3% sensitivity, 82.9% specificity, and 96% negative predictive value. Conclusion and Relevance A multimodal AI approach using routinely acquired chest computed tomography, echocardiography, and electrocardiography data can enable screening for ATTR-CM in TAVR patients, potentially facilitating earlier diagnosis and treatment initiation.
Building similarity graph...
Analyzing shared references across papers
Loading...
Nusrat Binta Nizam
Cornell University
Ashley Beecy
Sutter Health
Jacob Groenendyk
Cornell University
European Heart Journal - Digital Health
Cornell University
Columbia University Irving Medical Center
NewYork–Presbyterian Hospital
Building similarity graph...
Analyzing shared references across papers
Loading...
Nizam et al. (Tue,) conducted a observational in Transthyretin cardiac amyloidosis in severe aortic stenosis (n=816). Multimodal artificial intelligence (AI) model vs. Single-modality approaches was evaluated on Area under the receiver operating characteristic curve (AUROC) for ATTR-CM detection (AUROC 0.85, 95% CI 0.74-0.93). A multimodal AI model using chest CT, echocardiography, and ECG achieved an AUROC of 0.85 (95% CI 0.74-0.93) for detecting ATTR-CM in TAVR patients, outperforming single-modality approaches.
synapsesocial.com/papers/69e9b9e385696592c86ec696 — DOI: https://doi.org/10.1093/ehjdh/ztag064