<![CDATA[Diabetes is a major health problem that can significantly reduce life expectancy and increase the risk of serious complications such as kidney failure, stroke, and cardiovascular disease. Therefore, early detection is essential to prevent the progression of the disease. This study proposes a machine learning-based approach for early diabetes detection using a private dataset obtained from RSUP Persahabatan General Hospital in Jakarta, Indonesia. The dataset consists of 501 patient records with clinical and laboratory features extracted from the hospital’s electronic medical record system. Several machine learning algorithms were implemented and compared, including Logistic Regression, Support Vector Machine, Random Forest, Decision Tree, Naïve Bayes, Extreme Gradient Boosting, Ensemble methods, and Artificial Neural Networks. Feature selection was performed using ANOVA, and hyperparameter optimization was applied using GridSearchCV to improve model performance. The experimental results show that the Artificial Neural Network model achieved the best performance with an accuracy of 0.86 (86%). Statistical analysis using logistic regression identified systolic blood pressure, diastolic blood pressure, age, HDL cholesterol, and leukocyte levels as the most significant risk factors associated with diabetes. These findings demonstrate the potential of machine learning techniques to support early diabetes detection using clinical data from Indonesian healthcare settings.]]>
Copyrights © 2026
