Year: 2026
Type: article
Abstract: Artificial intelligence (AI) is increasingly being applied across obstetrics, with applications spanning electronic medical record analysis, prenatal ultrasound, fetal monitoring, prediction of pregnancy complications, labor management, postpartum care, and telemedicine. The evidence reviewed in this article indicates that AI can improve automated information extraction, image and signal interpretation, and risk discrimination across multiple obstetric tasks, with particularly consistent advances in fetal ultrasound analysis, cardiotocography (CTG) interpretation, and early prediction of gestational diabetes, preeclampsia, and preterm birth. However, model performance is heterogeneous and is often reported using discrimination or classification metrics rather than patient-centered clinical outcomes. In CTG research, models differ substantially in whether they classify interpretation categories, detect fetal hypoxia or acidemia, or predict neonatal outcomes, making direct comparison difficult. Similarly, high predictive accuracy for pregnancy complications does not establish that AI implementation itself improves maternal or neonatal outcomes, and the clinical benefit of treatment should not be attributed to the predictive model without intervention studies. Across obstetric applications, limited external validation, dataset shift, class imbalance, calibration, interpretability, workflow integration, and data governance remain major barriers to clinical translation. This review therefore emphasizes the distinction between technical model performance and demonstrated clinical utility, and highlights the evidence required to move AI applications from retrospective development toward prospective, clinically integrated obstetric care.
Source: Frontiers in Medicine
Institution: Qinghai Red Cross Hospital
Language: English
Subfield: Health Informatics
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $3,801
OpenAlex ID: W7225139845