Hybrid XGBoost-SVM Model untuk Sistem Pendukung Keputusan dalam Prediksi Penyakit Diabetes

Diabetes is a chronic disease that continues to rise globally each year, requiring early detection for more effective prevention. This study develops an artificial intelligence-based decision support system for diabetes prediction using a Hybrid XGBoost-SVM model. The model combines the Support Vector Machine (SVM), known for its interpretability, with XGBoost (XGB), which enhances accuracy through boosting techniques. The study utilizes the Pima Indians Diabetes Dataset, undergoing preprocessing, normalization, data splitting, and model training. The evaluation compares accuracy, precision, recall, and F1-score across the three models. Experimental results indicate that XGBoost and SVM both achieve an accuracy of 75%. However, the Hybrid XGBoost-SVM model provides consistently improved performance, achieving the highest accuracy (77%), along with increased precision (70%) and F1-score (65%). Although the numerical improvement in accuracy appears relatively small, this enhancement is significant in the medical context, especially due to improved precision and balanced classification. This study concludes that the Hybrid XGBoost-SVM approach offers a more optimal and reliable alternative in decision support systems for diabetes prediction. Future research can explore other model combinations, such as Stacking or Weighted Voting, to enhance predictive performance further.