ECG models identify concurrent high-risk imaging features in hypertrophic cardiomyopathy
Source check: Source checked. Not independent scientific confirmation.
Finding: A pretrained variational-autoencoder ECG representation supported classifiers for four imaging findings in hypertrophic cardiomyopathy. In233 patients at another centre, ROC areas ranged from0.68 for extensive enhancement to0.91 for apical aneurysm, with only one aneurysm-positive patient. This is retrospective reference-label evaluation, not a demonstration that imaging can safely be omitted.
Technology: Variational autoencoder; dense neural-network classifiers
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A VAE pretrained on88105 ECGs provides24 latent features to learned classifiers, alongside age, sex and heart rate, linking electrical patterns to contemporaneous imaging labels.
- Test and result
- Tufts1930 patients were partitioned1544 development and386 holdout, followed by233 Amrita patients. Endpoints were reduced ejection fraction, extreme wall thickness, apical aneurysm and extensive late gadolinium enhancement.
- Limitation / still unknown
- Only one external aneurysm-positive case makes that endpoint highly uncertain. CMR labels were available in selected1584/1930 development-site and104/233 external patients. Hypothetical imaging savings combined development and evaluation cases; they are not prospective savings or safety evidence. A linked editorial is authenticated as Comment/Editorial, not a correction; its full body was not reviewed. No clinical-benefit or causal interpretation is established.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC11272903/ · https://doi.org/10.1016/j.hrthm.2024.01.031
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Actual relevant primary computational/clinical Methods and high-level Results read; no operational biological protocols inspected.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.