Machine-Learning Model Predicts Preeclampsia in Late Pregnancy

Preeclampsia is a serious pregnancy complication marked by the sudden onset of high blood pressure before delivery. Affecting approximately 2% to 8% of pregnancies worldwide, the condition can lead to severe health risks for both the parent and the baby if not detected and managed promptly. While early screening tools exist, predicting cases that develop later in pregnancy remains challenging. Researchers have now developed a machine-learning model designed to estimate preeclampsia risk during the later stages of pregnancy.

The model, developed by researchers at Weill Cornell Medicine (New York, NY, USA), in collaboration with NewYork-Presbyterian (New York, NY, USA), analyzes electronic health record (EHR) data collected during the third trimester to generate continuously updated predictions of preeclampsia risk. Unlike earlier tools that provide a single early estimate, the system recalculates risk as new clinical information becomes available.

The research team trained the machine-learning model using data from 35,895 pregnancies at NewYork-Presbyterian/Weill Cornell Medical Center between October 2020 and May 2025. They then validated the model using additional datasets from 8,664 pregnancies at NewYork-Presbyterian Lower Manhattan Hospital and 14,280 pregnancies at NewYork-Presbyterian Brooklyn Methodist Hospital. The model showed its strongest predictive performance at approximately 34 weeks of pregnancy, providing clinicians with valuable lead time before delivery.

Among the most influential predictive factors were blood pressure measurements, which remained the strongest indicator of risk. Earlier in the third trimester, abnormal blood test results—potentially reflecting placental dysfunction—were also associated with increased risk. The study, published in JAMA Network Open, found that later in pregnancy, patient age and white blood cell counts became more important predictors, suggesting that inflammatory processes may contribute to the condition at this stage.

By continuously updating predictions using real-time clinical data, the model may help clinicians identify patients at elevated risk of preeclampsia during late pregnancy. Earlier detection could allow physicians to intensify monitoring, manage blood pressure more aggressively, and make informed decisions about delivery timing to protect maternal and fetal health. Researchers note that further studies are needed to better understand whether different forms of preeclampsia arise from distinct biological causes, such as placental dysfunction or systemic inflammation.

Related Links:
Weill Cornell Medicine
NewYork-Presbyterian