<![CDATA[Diabetes mellitus is a metabolic disease characterized by elevated blood sugar levels due to impaired insulin secretion, insulin action, or both. The disease has a major impact on public health and contributes to high morbidity and mortality rates in many countries. Prevention and early detection are essential to reduce the adverse effects of this disease. This study aims to analyze and apply machine learning algorithms in detecting diabetes mellitus, focusing on the use of XGBoost and LightGBM algorithms. The dataset used in this study includes various features related to diabetes risk factors, such as age, gender, body mass index (BMI), hypertension, smoking history, and HbA1c and blood glucose levels. Preprocessing was performed to clean and balance the data using the SMOTE-Tomek technique. Next, the model was built and evaluated using the K-Fold cross-validation method to measure the accuracy and stability of the model. The results showed that the XGBoost model achieved 97.31% accuracy, while the LightGBM model produced 97.26% accuracy. Combining the two models through blending techniques resulted in an accuracy of 97.51%, indicating that the combination of models can improve prediction performance. This study shows the great potential of machine learning algorithms, especially XGBoost and LightGBM, in detecting diabetes mellitus accurately and efficiently. Hopefully, the results of this study can contribute to the development of decision support systems for more effective early diagnosis of diabetes.]]>
Copyrights © 2025
