Preliminary diagnosis of pulmonary tuberculosis using ensemble method

Tuberculosis (TB) is the oldest human diseases with the highest mortality rates among infectious disease. Indonesia is the fifth highest TB burden country in the world. Diagnosis of TB is difficult, especially in the case of pediatric patients, extrapulmonary TB and smear-negative pulmonary TB. In addition, some of the tuberculosis symptoms have in common not only with lung cancer but also with other diseases. This situation will lead to a delay in the correct diagnosis and exposure to the inappropriate medication. Finally, individuals that receive inadequate treatment are more vulnerable to develop multidrug-resistant tuberculosis. This study aims to model the preliminary diagnosis of pulmonary tuberculosis. Preliminary diagnosis is established only by using patient demographic data, anamnesis, and physical examination. Some experiments were conducted using classification techniques. Some single classifiers such as C4.5, Naive Bayes, Backpropagation and SVM will be compared with ensemble method in order to improve the performance of the model. The data were taken from medical record of tuberculosis patients from Jakarta Respiratory Center. The result showed that ensemble method provided the best accuracy compared to the single classifier.