ISSN : 2319-7323

INTERNATIONAL JOURNAL OF COMPUTER SCIENCE ENGINEERING

ABSTRACT

Title	:	A Novel Approach For Detecting Tuberculosis Based On Observed Manifestations Using Supervised Machine Learning
Authors	:	Solomon Osarumwense Alile
Keywords	:	Tuberculosis; Mycobacterium Tuberculosis; Prediction; Detection; Artificial Intelligence, Supervised Machine Learning; Bayesian Belief Network.
Issue Date	:	Jul-Aug 2020
Abstract	:	Tuberculosis (TB) is a severe communicable respiratory tract infection caused by Mycobacterium Tuberculosis (MTB), thus infiltrating and disrupting the functionality of the lungs which is the major spot of infection. Tuberculosis is classified as a infectious disease transmitted via diminutive respiratory droplets let free into the air through coughs and sneezes from symptomatic patients which when inhaled by asymptomatic patients; it can be transmitted to these individuals inside the first few weeks, sometimes months leading to years before the manifestation of the infection begin to show in these infected patients with organs resident in their bodies such as kidneys, bladder and liver just name a few; being are at risk due to the spread of the infection. Owing to the widespread mode of TB infection, it was categorized as a global health pandemic by World Health Organization in 1993, which to date remains the world foremost infectious killer disease which has caused millions of untimely passings of infected persons. What’s more, the manifestations of TB are hemoptysis (cough with sputum covered with blood), angina (chest pains), breathlessness, weakness, weight loss, fever, night sweats and loss of appetite just to name but a couple. Nevertheless, in recent past, several systems have been developed to detect this transmittable ailment, yet they delivered a ton of bogus negative during testing and couldn't distinguish Tuberculosis in view of its covering symptoms it imparts to other Respiratory Tract Infections (RTIs). Consequently, there was the need to proffer a solution for the quagmire of under-diagnosis and misdiagnosis of Tuberculosis which is much uncontrolled in Sub-Sahara Africa and South-East Asia respectively. Hence, in this paper, we proposed and developed a model to predict Tuberculosis and Respiratory Tract Infections using an AI technique called Bayesian Belief Network. The model was structured using Bayes Server and tested with data retrieved from Tuberculosis machine learning repository. The model had an overall prediction exactness of 99.98%; 99.84% and 99.96% sensitivity of Tuberculosis and Respiratory Tract infections in that order.
Page(s)	:	264-280
ISSN	:	2319-7323
Source	:	Vol. 9, No. 4