Sex-related cardiovascular risk is non-dichotomous: artificial intelligence-enhanced electrocardiography reveals continuum of risk in females
Author(s)
Sau, A
Type
Journal Article
Abstract
Background Females are typically underserved in cardiovascular medicine. The use of sex as a dichotomous variable for risk stratification fails to capture the heterogeneity of risk within each sex. We aimed to develop an AI-ECG model to investigate sex specific cardiovascular risk. Methods We trained a convolutional neural network to classify sex using the 12-lead ECG. The Beth Israel Deaconess Medical Center (BIDMC) was the derivation cohort (1,163,401 ECGs). External validation was performed in the UK Biobank (UKB, 42,386 ECGs). We examined the difference between AI-ECG-predicted sex (continuous) and biological sex (dichotomous), termed “sex discordance score”. Findings AI-ECG accurately identified sex (BIDMC AUROC 0.943 (0.942-0.943), UKB AUROC 0.971 (0.969-0.972)). In BIDMC outpatients with normal ECGs, increased sex discordance score was associated with covariate-adjusted increased risk of cardiovascular death in females, but not males (Females HR 1.78 (1.18-2.70) p = 0.006, Males HR 1.00 (0.63-1.58), p = 0.996). In the UK Biobank cohort, the same pattern was seen (Females HR 1.33 (1.06-1.68) p = 0.015, Males HR 0.98 (0.80-1.20), p = 0.854). Females with a higher sex discordance score were more likely to have future heart failure or myocardial infarction and had more male cardiac (increased left ventricular mass, and chamber volumes) and non-cardiac phenotypes (increased muscle mass, reduced body fat percentage). Interpretation Sex discordance score is a novel AI-ECG biomarker capable of identifying females with disproportionately elevated cardiovascular risk. AI-ECG has the potential to identify female patients who may benefit from enhanced risk factor modification or surveillance. Funding British Heart Foundation
Date Acceptance
2024-12-06
ISSN
2589-7500
Publisher
Elsevier
Journal / Book Title
The Lancet: Digital Health
Copyright Statement
Subject to copyright. This paper is embargoed until publication. Once published the Version of Record (VoR) will be available on immediate open access.
License URI
Publication Status
Accepted
Rights Embargo Date
10000-01-01