A machine-learning exemplary developed by Weill Cornell Medicine investigators whitethorn supply clinicians pinch an early informing of a complication that tin hap precocious successful pregnancy.
Preeclampsia is simply a abrupt onset information that involves precocious humor unit anterior to delivery. It affects astir 2% to 8% of pregnancies worldwide and tin person superior consequences for some genitor and child. A caller study, published March 6 successful JAMA Network Open, describes a machine-learning-based machine exemplary that provides continually updated predictions of preeclampsia consequence based connected physics wellness grounds information recorded precocious successful pregnancy. The study was co-led by Dr. Fei Wang, subordinate dean for AI and information subject and nan Frances and John L. Loeb Professor of Medical Informatics successful Department of Population Health Sciences astatine Weill Cornell Medicine, and Dr. Zhen Zhao, professor of objective pathology and laboratory medicine astatine Weill Cornell Medicine and cardinal laboratory head astatine NewYork-Presbyterian/Weill Cornell Medical Center. Clinical expertise successful obstetrics was provided by Dr. Tracy Grossman, adjunct professor of objective obstetrics and gynecology astatine Weill Cornell Medicine and a maternal-fetal medicine master astatine NewYork-Presbyterian Brooklyn Methodist Hospital.
Existing models that measure preeclampsia consequence during nan first trimester are chiefly utilized arsenic early warnings, allowing clinicians to prescribe aspirin arsenic a preventive medicine early successful nan gestation and supply further monitoring passim at-risk pregnancies. While these approaches whitethorn trim nan consequence of early-onset preeclampsia, their predictive accuracy is constricted for late-onset and word cases, which relationship for nan mostly of preeclampsia diagnoses. As a result, fewer devices are disposable to thief foretell short-term preeclampsia consequence during nan past trimester of gestation erstwhile astir cases arise. To capable this gap, co-first authors Dr. Haoyang Li, a postdoctoral subordinate successful organization wellness sciences, and Dr. Yaxin Li, a postdoctoral subordinate successful pathology and laboratory medicine, worked pinch Drs. Wang, Zhao and Grossman to create and trial a preeclampsia modeling instrumentality utilizing deidentified physics wellness grounds information connected almost 59,000 pregnancies astatine 3 NewYork-Presbyterian hospitals. The squad created nan exemplary utilizing information connected 35,895 pregnancies of patients who delivered astatine NewYork-Presbyterian/Weill Cornell Medical Center betwixt October 2020 and May 2025. The exemplary astir accurately predicted nan likelihood of preeclampsia astir 34 weeks, perchance giving clinicians clip to return preventive measures.
The squad past validated their exemplary utilizing information from 8,664 pregnancies astatine NewYork-Presbyterian Lower Manhattan Hospital and 14,280 astatine NewYork-Presbyterian Brooklyn Methodist Hospital. The exemplary showed nan pregnant patient's humor unit was nan strongest predictor of preeclampsia. However, early successful nan 3rd trimester, abnormal results from regular testing of nan patient's humor whitethorn besides propose imaginable risk. These laboratory results whitethorn propose that emerging problems pinch nan placenta, which provides nutrients and oxygen to nan fetus, could beryllium contributing to preeclampsia astatine this stage. Later successful nan 3rd trimester, nan patient's property and achromatic humor compartment count became much important indicators, suggesting inflammation whitethorn beryllium playing a domiciled astatine this time.
The exemplary whitethorn thief clinicians place patients successful nan 3rd trimester of gestation astir apt to create preeclampsia and supply them further lead clip to return timely objective action, including enhanced monitoring, humor unit management, and decisions astir transportation timing. Unlike earlier approaches that supply a single, fixed consequence estimate, this exemplary continuously updates preeclampsia consequence pinch existent physics wellness grounds information arsenic gestation progresses, aligning prediction pinch real-world objective decision-making successful precocious pregnancy. More study is needed to find if preeclampsia astatine different stages of nan 3rd trimester has chopped causes, for illustration placental dysfunction aliases systemic inflammation. But if those patterns are confirmed, they whitethorn thief clinicians create much targeted preeclampsia interventions that reside nan guidelines causes.
Source:
Journal reference:
Li, H., et al. (2026). Machine Learning for Dynamic and Short-Term Prediction of Preeclampsia Using Routine Clinical Data. JAMA Network Open. DOI: 10.1001/jamanetworkopen.2026.0359. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2845997
English (US) ·
Indonesian (ID) ·