Evaluation of A Novel Risk Factor Screening Tool for Gestational Diabetes Mellitus: A Machine Learning Based Predictive Method

Introduction: Gestational diabetes mellitus (GDM) is a significant pregnancy complication linked to adverse outcomes for both mother and child. Early identification of high-risk individuals is crucial for effective management and prevention for the onset/progression of the GDM. Our study aims to a) evaluate the effectiveness of a newly developed machine learning based risk factor screening tool for predicting GDM and b) to compare its predictive performance against established models and current literature. Methods: This study explored SNP data from the leptin (LEP) and leptin receptor (LEPR) genes to develop machine learning models for predicting gestational diabetes mellitus (GDM). It included data preprocessing, such as cleaning and feature selection, focusing on genetic markers, metabolic parameters, and demographic information. Various algorithms, including Logistic Regression, Decision Trees, and Random Forests, were used, and their performance was evaluated using metrics like accuracy and ROC-AUC to determine the best model for GDM prediction. Results: The newly developed screening tool demonstrated a sensitivity of 85%, specificity of 78%, positive predictive value (PPV) of 68%, and negative predictive value (NPV) of 90% in predicting GDM. Comparatively, machine learning models showed higher sensitivity (90-95%) but lower specificity (65-75%). Conclusion: The developed risk factor screening tool is a viable method for predicting GDM, with accuracy metrics comparable to advanced machine learning models and established literature. Future research should focus on refining these tools and exploring their integration into routine prenatal care to enhance early detection and intervention strategies for GDM.