AI-Driven Non-Invasive Cardiac Risk Assessment in Women: A Technical Review and Conceptual Framework

Cardiovascular disease continues to be the leading cause of mortality worldwide, yet itsmanifestation in women remains insufficiently understood and frequently under diagnosed[1,5]. A particularly challenging aspect lies in the interpretation of mild or ambiguous chestdiscomfort, which often leads to anxiety-driven hospital visits or, conversely, dangerousdelays in seeking care. This paper explores the possibility of addressing this gap through anon-invasive, AI-assisted screening approach designed specifically with women’s symptomprofiles in mind. Rather than proposing a purely diagnostic tool, this work positions itself as aconceptual and technical bridge between symptom awareness and clinical intervention.The paper reviews existing artificial intelligence applications in cardiovascular risk predictionand identifies a critical limitation: the lack of integration of atypical symptoms andpsychological factors such as anxiety [6,7]. Building on this observation, a structuredframework is proposed that combines symptom inputs, demographic attributes, and optionalphysiological signals within a multimodal learning architecture. Particular emphasis is placedon interpretability, with SHAP and LIME discussed as mechanisms to ensure that modeloutputs remain transparent and clinically meaningful [3,4].While the framework is not experimentally validated within this study, it is designed withpractical deployment considerations in mind. The discussion outlines a phased validationpathway, suggesting how such a system could be incrementally tested and refined in realworld settings. Overall, the work aims to contribute not just a technical perspective, but aclinically sensitive approach to early-stage cardiac risk awareness in women.