From Data to Diagnosis: The Role of Machine Learning in Preeclampsia Prediction

Preeclampsia (PE) is a complex, potentially life threatening hypertensive disorder of pregnancy, that affects approximately 2-5% of pregnancies worldwide. It is characterised by elevated Blood Pressure (BP), Proteinuria and multiorgan dysfunction, posing significant risks to both maternal and foetal health. Timely identification of high-risk pregnancies is crucial to prevent severe complications. However, conventional diagnostic approaches relying on clinical assessments and biomarkers analysis often fall short in early and accurate prediction. Recent advancements in Machine Learning (ML) have opened new avenues for improving early prediction and diagnosis of PE, offering the potential for more targeted and timely interventions. This review explores the integration of ML algorithms with clinical, laboratory and demographic data to develop robust predictive models for PE. Studies employing Support Vector Machines (SVM), Logistic Regression (LR), Random Forest (RF) and Deep Neural Networks (DNN) has demonstrated the ability to uncover complex non linear patterns associated with PE risk. These models often surpass traditional diagnostic methods in accuracy and offer transformative potential in prenatal care. By incorporating ML based tools into routine obstetrics practice, it can help healthcare providers predict PE early and implement preventive strategies to reduce the risk of complications such as eclampsia and organ failure. This review offers a comprehensive, evidence-based review of ML models in PE prediction by integrating clinical, biochemical, imaging and genomic data. By bridging obstetrics and data sciences, it highlights model performance, personalisation, and clinical relevance, making it a timely and translational resource for improving maternalfoetal outcomes through AI driven healthcare.