Implementation of Logistic Regression for Classification of Chronic Kidney Disease with and without Comorbidities: A Performance Evaluation

Chronic kidney disease (CKD) is a highly progressive condition that can lead to kidney failure and requires early detection to prevent serious complications. This study analyzes the application of Logistic Regression algorithm in the classification of CKD with and without comorbidities, using medical record data of outpatients at RSUD Al-Ihsan Bandung in 2024. The dataset includes clinical attributes such as age, gender, ureum, creatinine, blood pressure, blood sugar level, hemoglobin, leukocytes, and erythrocytes. The classification process was performed through Grid Search and k-fold cross-validation techniques to find the best hyperparameter combination. The highest accuracy of 80% was obtained at a data split of 90:10, with optimal configurations including liblinear solver, penalty l1, and class weight balanced. The classification report showed precision of 0.79, recall of 0.85, and F1-score of 0.82 for class 0 (CKD), and precision of 0.80, recall of 0.72, and F1-score of 0.76 for class 1 (CKD with comorbidities). The AUC value of 0.80 on the ROC curve indicates good discriminatory ability in distinguishing the two classes. This study proves the potential use of machine learning algorithms in supporting early detection of CKD.