Full bibliography

Background: Maternal mortality remains a major global health challenge, disproportionately affecting black and Indigenous women. Hypertensive disorders of pregnancy and postpartum hemorrhage are the leading direct causes of maternal death. Artificial intelligence (AI) tools have emerged as potential strategies for predicting these complications, yet concerns persist about their equity and validation across racial groups. Methods: A rapid review was conducted in five databases, PubMed, EMBASE, Web of Science, Scopus and LILACS, to synthesize recent evidence on the use of AI for preventing maternal mortality due to hypertension and postpartum hemorrhage. Studies published in the last five years that included racial or ethnic data were selected and analyzed narratively. Results: Ten studies met the inclusion criteria, showing high predictive accuracy of AI models (AUROC often >0.95) for severe maternal outcomes. However, few models incorporated racial variables or underwent external validation in racially diverse or low-resource populations. Evidence suggests that unrepresentative datasets may perpetuate or exacerbate existing health inequities. Conclusions: AI demonstrates strong technical performance in predicting maternal complications but limited equity in application. Broader racial representation, external validation, and ethical governance are essential for ensuring that AI-based tools reduce rather than reinforce racial disparities in maternal mortality.