AI-Enabled Early Detection of Preeclampsia: A Predictive Model Based on Multivariate Biomarker Analysis.

Abstract

Preeclampsia remains one of the leading causes of maternal and perinatal morbidity worldwide, characterized by hypertension and systemic organ dysfunction during pregnancy. Early detection is essential to prevent adverse outcomes, yet current clinical approaches largely rely on symptomatic presentation rather than predictive insight. This study proposes an AI-enabled predictive model that integrates multivariate biomarker data, clinical parameters, and maternal demographics to identify high-risk cases of preeclampsia in the early stages of gestation. Using a dataset of 2,500 pregnant women, key biomarkers such as sFlt-1, PlGF, and uric acid levels were analyzed alongside systolic and diastolic blood pressure, BMI, and gestational age. A hybrid ensemble model combining Random Forest, Support Vector Machine (SVM), and Gradient Boosting algorithms achieved an accuracy of 94.8% and an AUC of 0.96 in early-stage prediction. Correlation analysis revealed that the ratio of sFlt-1/PlGF was the most significant predictor, followed by blood pressure variability. The results underscore the potential of AI in transforming prenatal diagnostics by enabling proactive interventions and improving maternal health outcomes. The model provides a scalable, data-driven approach that can be integrated into digital prenatal care systems for real-time clinical decision support.