Detection of Diabetes in Pregnant Women Using Machine Learning as an Effort Towards Golden Indonesia 2045

Abstract

One of the goals of the Golden Indonesia 2045 program is to utilize health technology to enhance public health, with diabetes being a prominent concern. This research aims to employ ensemble classifier optimization techniques in machine learning for the early detection of diabetes among pregnant women. The study uses physiological data, including variables such as glucose levels, number of pregnancies, skin thickness, blood pressure, insulin levels, body weight, family history, and age. By combining multiple models, ensemble classifiers can enhance prediction accuracy, stability, and overall model performance. This research utilizes an open Kaggle dataset on pregnant women to train and test machine learning models, specifically Support Vector Machine (SVM) and Deep Learning, incorporating ensemble techniques such as bagging and boosting. Experimental results indicate that the ensemble classifier approach significantly enhances diabetes classification, with SVM using bagging achieving the highest accuracy at 76.95%. These findings suggest that ensemble classifier methods could be a valuable tool for early diabetes detection, providing timely intervention and improved risk management during pregnancy, which supports the objectives of improving public health under the Golden Indonesia 2045 initiative.