<![CDATA[Tuberculosis (TB) remains a global health challenge, particularly in developing countries such as Indonesia, which ranks third worldwide in the number of TB cases. This study aims to evaluate the performance of Support Vector Machine (SVM) and Random Forest (RF) algorithms in predicting TB patient recovery rates based on clinical data obtained from healthcare facilities in Indonesia. Evaluation results indicate that the model achieved very high precision scores (100%) for the "Deceased," "Transferred," and "Default" categories; however, these findings require critical interpretation due to the likely class imbalance in those categories. In contrast, for the "Recovered" and "Completed" categories—where data instances were more numerous—the model exhibited lower precision and recall values (below 90%), reflecting challenges in accurately predicting majority classes. These results suggest that despite seemingly high numerical performance, model predictions can be biased if class distribution is not appropriately considered. The main contribution of this research lies in providing a comparative analysis of two widely used machine learning algorithms in predicting TB recovery outcomes, while emphasizing the importance of addressing data imbalance issues in clinical predictive modeling. The findings provide a practical basis for integrating predictive algorithms into clinical workflows, enabling more accurate monitoring of patient recovery and timely adjustments of TB treatment plans in Indonesia.]]>
