Bashir, Sharmake Gaiye, Salad, Hiba Abdi, Abdullahi, Yakub Burhan, Abdi, Yusuf Hared, Abdi, Mohamed Sharif, Ahmed, Naima Ibrahim, Saidu Musa, Shuaibu, Elehamer, Nafisa M. K., Musa, Muhammad Kabir, Bolarinwa, Obasanjo ORCID: https://orcid.org/0000-0002-9208-6408 and Dada, Olusegun (2026) Artificial intelligence for predicting and preventing adverse pregnancy outcomes addressing bias and clinical translation. Frontiers in Digital Health, 8. p. 1841706.

Preview

Text fdgth-8-1841706.pdf - Published Version Available under License Creative Commons Attribution. Download (822kB) | Preview

Abstract

Artificial intelligence (AI) has emerged as a promising approach for improving the early detection and management of adverse pregnancy outcomes through enhanced risk prediction and clinical decision support. This narrative review synthesizes current evidence on AI applications for predicting major obstetric complications, including preeclampsia, preterm birth, gestational diabetes, and fetal growth restriction. Reported predictive performance across studies demonstrates considerable heterogeneity, with area under the receiver operating characteristic curve (AUROC) values ranging from approximately 0.73 to 0.97, reflecting differences in datasets, model architectures, and validation strategies. Beyond predictive accuracy, this review critically examines sources of algorithmic bias that may influence model performance and equity in maternal healthcare. Eight key bias mechanisms are identified, including sampling bias, measurement bias, algorithmic bias, temporal bias, selection bias, labelling bias, deployment context bias, and access bias. These biases may limit model generalizability and risk amplifying existing maternal health disparities, particularly in low- and middle-income countries. Current evidence is further constrained by limited external validation across diverse populations, the absence of prospective clinical impact trials, insufficient cost-effectiveness analyses, and evolving regulatory frameworks governing AI accountability. The review discusses potential pathways for responsible clinical translation, emphasizing inclusive dataset development, rigorous multisite validation, careful integration into clinical workflows with human oversight, and strengthening regulatory and workforce capacity. Achieving equitable implementation of AI in maternal health will require deliberate efforts to embed transparency, accountability, and health equity throughout the AI development and deployment lifecycle.

Item Type:	Article
Status:	Published
DOI:	10.3389/fdgth.2026.1841706
School/Department:	London Campus
URI:	https://ray.yorksj.ac.uk/id/eprint/15319

University Staff: Request a correction | RaY Editors: Update this record