<![CDATA[Chronic liver disease is one of the leading causes of global morbidity and mortality, including in Indonesia. Early detection is essential to prevent disease progression to cirrhosis and hepatocellular carcinoma; however, clinical diagnosis is often delayed due to non-specific early symptoms and limited access to invasive diagnostic procedures. This study aims to develop and optimize a machine learning–based predictive model for early detection of chronic liver disease using clinical patient data. The Liver Cirrhosis dataset obtained from Kaggle was utilized, with preprocessing steps including missing value imputation, categorical variable encoding, feature selection using SelectKBest, and class imbalance handling through the Synthetic Minority Oversampling Technique (SMOTE). Three classification algorithms—Support Vector Machine (SVM), Random Forest, and XGBoost—were evaluated under a binary classification scheme of early-stage and advanced-stage disease. Model performance was assessed using accuracy, precision, recall, F1-score, and AUC-ROC metrics. The results indicate that the integration of feature selection and SMOTE improves model performance, particularly in enhancing sensitivity toward advanced-stage cases. XGBoost achieved the best overall performance based on AUC-ROC values, while Random Forest demonstrated a favorable balance between predictive performance and computational efficiency. This approach shows strong potential as a clinical decision support tool for early screening of chronic liver disease.]]>
