<![CDATA[General Background: Preeclampsia is a leading cause of maternal and perinatal morbidity and mortality worldwide, with early prediction remaining a critical challenge in obstetric care. Specific Background: Conventional diagnostic approaches based on clinical and isolated biochemical markers often identify the disorder at advanced stages and fail to capture its multifactorial pathophysiology. Knowledge Gap: There is limited integration of multidimensional clinical, biochemical, and Doppler data into robust predictive models capable of early and individualized risk assessment. Aims: This study aimed to develop and evaluate a machine learning–based model for early prediction of preeclampsia using comprehensive antenatal data. Results: In a retrospective cohort of 1,200 pregnant women, the Extreme Gradient Boosting (XGBoost) model demonstrated superior performance, achieving an AUC of 0.94, sensitivity of 91%, specificity of 89%, and overall accuracy of 90%, outperforming random forest, support vector machine, and logistic regression models. Key predictors included mean arterial pressure, maternal age, uterine artery pulsatility index, placental growth factor, and soluble fms-like tyrosine kinase-1. Novelty: The study integrates 35 heterogeneous parameters into an AI-driven framework, highlighting the strength of ensemble learning in capturing nonlinear risk patterns. Implications: AI-based predictive tools offer significant potential for early identification of high-risk pregnancies, enabling targeted preventive interventions and advancing precision obstetrics to reduce preeclampsia-related adverse outcomes.Highlight : XGBoost showed high accuracy for early preeclampsia risk prediction. Combined clinical, biochemical, and Doppler data enabled early risk identification. Early prediction supports timely preventive obstetric interventions. Keywords : Preeclampsia, Pregnancy, Prediction, Artificial Intelligence, Machine Learning]]>
