A new machine-learning model developed by researchers at Weill Cornell Medicine may help doctors detect preeclampsia risk in pregnant patients during the later stages of pregnancy. Preeclampsia, a condition marked by sudden high blood pressure before delivery, affects between 2% and 8% of pregnancies worldwide and can lead to serious health problems for both parent and child.
The study, published March 6 in JAMA Network Open, details how the computer model uses electronic health record data collected late in pregnancy to make ongoing predictions about preeclampsia risk. The research was led by Dr. Fei Wang, associate dean for AI and data science at Weill Cornell Medicine; Dr. Zhen Zhao, professor of clinical pathology and laboratory medicine at Weill Cornell Medicine; and Dr. Tracy Grossman, assistant professor of clinical obstetrics and gynecology at Weill Cornell Medicine.
Current models that assess preeclampsia risk are mostly used early in pregnancy, often leading to preventive measures such as prescribing aspirin or increasing monitoring for those considered at higher risk. However, these tools have limited accuracy when it comes to predicting cases that arise later in pregnancy—the period when most diagnoses occur.
To address this gap, co-first authors Dr. Haoyang Li and Dr. Yaxin Li worked with their colleagues to develop a predictive tool using deidentified electronic health records from nearly 59,000 pregnancies across three NewYork-Presbyterian hospitals. The model was built using data from over 35,000 pregnancies at NewYork-Presbyterian/Weill Cornell Medical Center between October 2020 and May 2025. According to the study's findings, the model is most accurate around week 34 of gestation—potentially providing clinicians with time to intervene if needed.
Researchers validated the model with additional data from more than 8,600 pregnancies at NewYork-Presbyterian Lower Manhattan Hospital and over 14,000 at NewYork-Presbyterian Brooklyn Methodist Hospital. They found that blood pressure readings were the strongest indicator of preeclampsia risk overall. In early third trimester cases, abnormal results from routine blood tests also signaled increased risk—possibly reflecting placental issues affecting fetal development. Later in the third trimester, factors such as patient age and white blood cell count became more important predictors.
The team suggests that this approach could allow healthcare providers to identify patients who are most likely to develop preeclampsia during the final months of pregnancy and give them more time for clinical interventions like enhanced monitoring or adjusting delivery plans.
"Unlike earlier approaches that provide a single, static risk estimate," said the researchers in their report, "this model continuously updates preeclampsia risk with current electronic health record data as pregnancy progresses, aligning prediction with real-world clinical decision-making in late pregnancy."
Further research is needed to determine whether different causes underlie preeclampsia depending on when it develops during the third trimester—for example placental dysfunction versus systemic inflammation—but identifying such patterns could eventually lead to more targeted treatments.
