Perbandingan Metode Naïve Bayes dan Random Forest dalam Memprediksi Penyakit Diabetes Melitus pada Klinik Citra Sejati

Abstrak

Diabetes mellitus is a chronic disease with a steadily increasing prevalence in Indonesia and is one of the leading causes of death, particularly in urban areas. Early detection of this disease is crucial to prevent serious complications such as heart disease, kidney failure, and vision impairment. In the era of digital transformation, machine learning techniques offer great potential to support early and automated diagnosis with higher accuracy. This study aims to develop a diabetes prediction system based on medical record data using two machine learning algorithms: Naïve Bayes and Random Forest. The dataset was obtained from Klinik Citra Sejati, consisting of 266 patient records with seven clinical features: age, gender, leukocytes, platelets, hematocrit, erythrocytes, and erythrocyte sedimentation rate (ESR). The models were implemented using Python programming language and the Scikit-learn library. Performance evaluation was carried out using the confusion matrix and classification metrics such as accuracy, precision, recall, and F1-score. Furthermore, ROC curve analysis and 95% confidence interval calculation were used to assess the stability and reliability of the predictions. The results showed that the Random Forest algorithm achieved an average accuracy of 89.97% with an AUC of 0.93, while Naïve Bayes achieved an accuracy of 85.97% with an AUC of 0.72. Based on these results, Random Forest is considered more effective for diabetes classification and is recommended as the primary algorithm for the development of clinical decision support systems based on local medical data.