[Gender bias and artificial intelligence in cardiology: evidence, clinical implications, and future perspectives]

Artificial intelligence is increasingly used in cardiovascular medicine, with applications in diagnosis, risk prediction, and clinical decision-making. However, emerging evidence suggests that these tools may exhibit differences in performance between men and women, with the potential to amplify existing disparities in cardiovascular care. This narrative review examines the role of sex and gender in artificial intelligence models applied to cardiology, focusing on the main areas of application, including electrocardiography, cardiovascular imaging, risk prediction, and clinical decision support systems. In many settings, models demonstrate good overall performance; however, stratified analyses reveal reduced sensitivity and higher rates of false-negative results in women. These differences may lead to underdiagnosis, delayed treatment, and inadequate risk stratification. Major contributing factors include the underrepresentation of women in training datasets, the lack of sex-specific variables, and the use of non-sex-specific diagnostic criteria. A systematic evaluation of model performance across subgroups, together with the integration of sex- and gender-specific variables, is essential to ensure a more equitable and clinically appropriate use of artificial intelligence in cardiovascular practice.