An Identification of Tuberculosis (Tb) Disease in Humans using Naïve Bayesian Method

Abstract

Tuberculosis (TB) is a disease that can cause a death if not recognized or not treated properly. To reduce the death rate of tuberculosis patients, the health experts need to diagnose that disease as early as possible. Based on the main indication data, laboratory test results and the  rontgen photo, Naïve Bayesian approach in data mining techniques could be optimized to diagnose tuberculosis. Naïve Bayes classifiers predict class membership probabilities with a class that has the highest probability value. The output of the system is an identification Tuberculosis type of the patients. Testing of the system using 237 data sample with variation of cross-validation in 3, 5, 7 and 9-fold cross validation gives an average accuracy 85,95%.