<![CDATA[Chronic Kidney Disease (CKD) is one of the global health problems with increasing prevalence and mortality rates, requiring accurate early detection methods to support timely treatment and prevention. Various previous studies have applied machine learning techniques for CKD prediction; however, most studies are still limited to basic model implementation without systematic parameter optimization or real-time web-based prediction system deployment. This study aims to develop a CKD prediction model using the Support Vector Machine (SVM) algorithm optimized through GridSearchCV to improve classification performance. The research was conducted based on the CRISP-DM framework using the CKD dataset from the UCI Machine Learning Repository. The preprocessing stage included categorical data transformation, missing value handling using median imputation, and feature standardization using StandardScaler. Parameter optimization was performed by testing several SVM parameter combinations using a 5-fold cross-validation approach. The results showed that the optimized SVM model achieved an accuracy of 98.75%, with high precision, recall, and F1-score values in CKD and non-CKD classification. These results indicate better performance compared to several previous studies using similar datasets with accuracy below 98%. Furthermore, the model was implemented in a web-based application using Gradio and Hugging Face Spaces to support real-time prediction. Initial validation by an internal medicine specialist indicated that the system predictions were consistent with medical interpretation, suggesting that the proposed model has potential as a decision-support tool for early CKD detection]]>
