Classification of Pulmonary Tuberculosis using Mathematical Modeling and Machine Learning

Many research works have been conducted and developed to diagnose pulmonary Tuberculosis in recent years. There are thousands of common symptoms for several diseases, making it very difficult to find which symptoms are responsible for pulmonary Tuberculosis. In this paper, we have proposed a comparative analysis between local covering technique and the support confidence method to generate an association rule that gives correct information regarding symptoms of pulmonary Tuberculosis. We have validated our claim using the chi-square test. Then K- Nearest Neighbor (KNN), Support Vector Machines (SVM), and Logistic Regression were used to classify the presence or absence of pulmonary tuberculosis in the chest x-ray dataset which is available at Kaggle.