The role of Artificial Intelligence in Predicting Pregnancy Complications

Abstract

Pregnancy carries a risk of complications that can affect both mother and fetus, and many of these share a common origin in impaired placental development. Because they develop silently long before symptoms appear, early prediction remains a major challenge, and current risk-factor-based screening fails to identify many women at risk. This narrative literature review examines how artificial intelligence (AI), including machine learning and deep learning, can analyse complex clinical, biochemical and imaging data to predict preeclampsia, fetal growth restriction, preterm birth and preterm premature rupture of the membranes, and compares it with conventional screening strategies. Current evidence suggests AI may improve prediction, particularly with complex data such as ultrasound images, but it has not consistently demonstrated an advantage over well-established statistical models on structured clinical data. Important barriers to clinical implementation include reliance on retrospective data, limited external validation, the "black box" problem, and concerns about the privacy of highly sensitive maternal–fetal data. AI therefore appears most valuable as a complementary tool to existing screening, and prospective studies measuring maternal and perinatal outcomes must confirm its clinical benefit.