Sex-specific assumptions underlie cardiovascular digital twin technologies: A narrative review.

Digital twin technologies (DTTs) are increasingly applied in cardiovascular medicine to support personalized treatment. At the same time, growing evidence demonstrates sex differences in cardiovascular anatomy, physiology, disease presentation, and outcomes. Whether current cardiovascular DTTs adequately incorporate these sex-specific characteristics is unclear. This narrative review examines how sex bias and inclusivity are addressed within cardiovascular DTTs and identifies where sex-related bias may arise within different components of DTT. A six-dimensional digital twin framework is introduced to review DTTs. We focus on three cardiovascular domains: coronary artery disease, aortic valve stenosis, and atrial fibrillation. For each domain, we assessed sex representation in modeling assumptions, data interpretation, clinical outputs, and validation studies. Within three domains, physiological assumptions, boundary conditions, interpretation thresholds, and validation cohorts, models are frequently derived from sex-skewed populations. DTTs estimating noninvasive fractional flow reserve are predominantly validated in male-dominated cohorts, potentially resulting in lower precision in women. In contrast, validation studies for DTTs in TAVI planning show variable sex distributions, with some cohorts being female-skewed and others male-skewed, raising questions about their generalizability across sexes. In both model-based DTTs, sex bias may arise from generalized boundary conditions. Electro-anatomical mapping systems for atrial fibrillation are susceptible to sex bias within measurement methodology. Uniform clinical thresholding and outcome selection bias further add to sex-bias in DTTs. Current cardiovascular DTTs insufficiently account for sex-specific cardiovascular characteristics, risking underperformance in underrepresented populations. Incorporating sex-aware physiological parameters, sex-stratified validation, balanced datasets, and transparent reporting should be considered minimal standards for future clinical implementation.