Application of Random Forest Method to Analyze the Effect of Smoking History on The Type and Outcomes of TB Examinations

Tuberculosis (TB) continues to pose a major global health challenge, especially in developing countries. One of the key risk factors that exacerbates the condition of TB patients is smoking, which increases susceptibility to infections and worsens disease prognosis. This study aims to evaluate the influence of smoking history on the type and outcomes of TB diagnoses using a Random Forest machine learning model. The dataset comprises information from TB-diagnosed patients, including demographic details such as age, gender, smoking status, patient type, and diagnostic results. The Random Forest model achieved an accuracy of 87.36%, performing best in classifying non-TB-infected patients. However, the model struggled to accurately identify healthy individuals without TB, likely due to data imbalance. This research offers fresh insights into the potential of machine learning to enhance TB diagnosis and prevention, while deepening the understanding of smoking as a risk factor in TB management.