Author(s): Prashant Kaler and Swaroopa Shastri

Abstract: Gestational Diabetes Mellitus (GDM) is a type of diabetes that develops during pregnancy and, if left untreated, can lead to serious complications for both mother and baby. Early detection is essential to reduce health risks, yet traditional screening methods such as the Oral Glucose Tolerance Test (OGTT) are usually performed late in pregnancy, limiting the time for preventive care. This study explores the use of machine learning (ML) techniques to predict the risk of GDM using clinical and demographic data. We used the well-known Pima Indians Diabetes Dataset, which includes eight key features: number of pregnancies, plasma glucose concentration, diastolic blood pressure, skin thickness, serum insulin, body mass index (BMI), diabetes pedigree function, and maternal age. Nine different ML algorithms—Logistic Regression, Random Forest, Decision Tree, Gradient Boosting, AdaBoost, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Naive Bayes, and XGBoost—were developed and tested. The dataset was divided into 70% training and 30% testing sets, and model performance was evaluated using Accuracy, Precision, Recall, and F1-score. Among these methods, the Random Forest model achieved the best overall results with an accuracy of 79% and a balanced F1-score of 0.66. Feature-importance analysis identified glucose level, insulin concentration, BMI, and number of pregnancies as the most significant predictors of GDM. These findings demonstrate that machine learning can be a valuable tool for early risk prediction, helping healthcare providers and expectant mothers take preventive steps before conventional screening is possible.

DOI: 10.33545/26633582.2025.v7.i2b.211

Pages: 152-159 | Views: 317 | Downloads: 100

Download Full Article: Click Here