Tuberculosis Analysis by Naive Bayesian Classification

Abstract

Tuberculosis (TB) is a social disease with medical aspects. By the increasing availability of biomedical and health-care data with a wide range of characteristics, computer-based medical system is playing an increasingly relevant role in assisting both diagnosis and treatment. Base on the knowledge stored, the system will learn the patterns using Naive Bayesian classification and decides the category of TB by probabilities. It is based on the theorem of posterior probability. This system intends to develop a diagnosis system of automatic classification method for TB diagnosis based on the symptoms of the patients. The system stores the knowledge of the medical experts and the medical records of the previous case as Training database. This system also considers the missing value by filling data completely, because it needs the actual symptoms of the patient for increasing accuracy to classify. This system can give the category of TB and treatment for the patient who has TB symptoms by using Naive Bayesian classification method on the Training database. The accuracy of the system for that patient is shown by using hold-out method on testing database.