Abstract / Summary
ynecologic and maternal health combine a substantial burden with care pathways in which delayed information can cause harm. In 2023, approximately 260,000 maternal deaths occurred worldwide, about 92% in low- and lower-middle-income countries; in 2024, cervical cancer accounted for approximately 604,000 new cases and 280,000 deaths. AI should not be presented as a substitute for obstetric, gynecologic, or public-health judgment. The most mature evidence concerns cervical screening and image analysis, where AI may support cytology, colposcopy, prioritization, and quality control. Fetal ultrasound and fetal echocardiography are promising for standard-plane acquisition, measurement, structure detection, and interpretation support. Risk models for preeclampsia, gestational diabetes, and perinatal complications are numerous, but heterogeneity, risk of bias, limited external validation, and differences in available variables restrict clinical transfer. Two Mexican cohorts demonstrate regional capacity to develop and test models for pre-eclampsia and gestational diabetes. They do not establish prospective clinical utility, improved outcomes, routine implementation, or transferability across Latin America. Responsible regional evaluation therefore requires external validation, defined clinical actions, bias audits, privacy safeguards, human review, and process measures before claims of clinical impact. This is a narrative community whitepaper, not a formal scoping review or an adoption recommendation.