Artificial Intelligence-based analysis of pre-eclampsia gene expression profiles for identification of novel potential pre-eclampsia diagnostic biomarkers.

Preeclampsia (PE) is a multifactorial and heterogeneous hypertensive disorder of pregnancy that poses significant diagnostic and therapeutic challenges. Identifying robust and generalizable biomarkers is critical for early detection and improved clinical outcomes. In this study, we present a machine learning-based framework for the analysis of gene expression profiles to uncover potential diagnostic biomarkers for PE. Multiple publicly available RNA-seq datasets were integrated, and differentially expressed genes (DEGs) were identified to construct a filtered gene subset. Using Recursive Feature Elimination (RFE), we evaluated optimal gene signatures comprising 20 to 50 features. Among several classifiers tested, XGBoost consistently outperformed others in accurately classifying PE samples. Notably, the most predictive gene sets revealed enrichment in zinc finger protein families, regulators of leptin signaling, and proteins involved in uterine vascular remodeling via G-protein-coupled estrogen receptors. These findings provide new insights into the molecular mechanisms underlying PE and highlight promising candidate biomarkers for diagnostic model development. Our approach underscores the potential of machine learning to advance personalized and early diagnosis of complex pregnancy disorders.