Predicting Gestational Diabetes Using GNN A GraphTheoretic Perspective on Patient Data

This study presents a novel approach to predict gestational diabetes using Graph Neural Networks (GNNs) and graph theory. The dataset, sourced from a medical database, includes features like age, pregnancies, BMI, HDL levels, and blood pressure. Data preprocessing involves filling missing values with column means and standardizing features. A graph is constructed with patients as nodes, each characterized by their features and class labels (GDM or Non-GDM). Edges represent significant relationships based on Euclidean distance. Visualization using a spring layout reveals clusters of similar patients, enhancing understanding of gestational diabetes predictors and potentially improving early diagnosis and patient management