Women have long been inadequately represented in cardiovascular medicine, as conventional risk evaluations fail to account for significant differences between the sexes. A team of researchers developed an AI-enhanced electrocardiography (AI-ECG) model to detect nuanced sex-specific cardiovascular risks. By analyzing patterns in over a million ECGs, the model introduces a novel biomarker, the “sex discordance score,” which highlights elevated risks of heart disease and related conditions in women. This study, published in The Lancet Digital Health, could pave the way for more precise and proactive care tailored to women’s unique cardiovascular needs.
It’s not all binary
Historically, women have been underdiagnosed and underserved in preventive care due to a perceived lower risk of heart disease. While traditional risk models treat sex as a binary variable, growing evidence suggests that cardiovascular risks may lie along a continuum rather than a strict male-female divide. Leveraging artificial intelligence, cutting-edge ECG models not only identify sex but also uncover “misclassifications” linked to adverse outcomes—raising the possibility that AI is detecting a deeper, continuum-based biological reality. This approach could redefine risk assessments and revolutionize precision medicine for cardiovascular care.
Imperial College London researchers analyzed over 1.2 million ECGs from a diverse cohort, finding that AI could predict biological sex with remarkable accuracy by identifying specific ECG features, such as ventricular depolarization time (QRS duration), ventricular repolarization characteristics (T-wave morphology), and heart rate. Arunashis Sau, PhD, and colleagues found that when discrepancies arose between AI-predicted and biological sex—termed sex discordance—women with higher “sex discordance scores” exhibited significantly increased risks of cardiovascular events like heart failure and myocardial infarction. The researchers validated the AI-generated “sex discordance scores” with data from the UK Biobank on over 500,000 individuals.
Through advanced phenome- and genome-wide analyses, researchers found that women with higher sex discordance scores tended to display traits traditionally associated with male cardiovascular phenotypes, including larger heart size and reduced fat mass. Genetic analyses suggested that sex discordance scores might be influenced by genetic variants linked to heart structure and hormonal pathways, further cementing the biomarker’s physiological and genetic underpinnings.
Sex-based cardiovascular risks
These findings suggest that AI-enhanced ECGs could transform risk assessment and personalize care by identifying subtle, sex-specific health vulnerabilities before they manifest as diseases.
While the study highlighted some limitations, such as its focus on self-identified sex and a lack of data on hormone levels and life events like pregnancy or menopause, its implications are profound. By integrating AI-driven insights into clinical practice, healthcare systems could better address the inequities in cardiovascular care and improve outcomes for women—a population often overlooked in traditional paradigms. The researchers propose embedding this biomarker into electronic health records to personalize preventive care, paving the way for a more equitable and precise approach to cardiovascular health.
