Pattern Identification of Drug Resistance for Tuberculosis using Machine Learning Techniques

Abstract

Among all other infectious diseases the 2^nd most infectious disease is the Tuberculosis (TB) which is caused by a gram positive bacterium called Mycobacterium tuberculosis, it is the main cause of deaths in the people of the developing countries. There are different types of Tb in some people the disease progression is immediate and is called active TB but in some patients the bacteria remain in the inactive state and is called inactive TB. Multiple drugs are being used for the treatment of the TB but the people are becoming resistant to these drugs because every human being has a different response towards the drugs. In the health care domains, data mining used for the processing of the massive data. The aims and objectives of this research were to explain some supervised learning algorithms used in finding the drugs resistance in the patients of different regions of Pakistan and also the type of the drug resistance developed in the patients of Pakistan. These algorithms include the random forest algorithms, decision tree, Naïve Bayes classifier and the support learning machine. The results showed that the highest accuracy of results were achieved by the use of the Naïve Bayes classifier which has given 96.5% accuracy. It was concluded from the research that this research can be expanded by applying other different data mining techniques like the clustering and the time series.