Diagnosis of Pulmonary Tuberculosis Using Naive Bayes Algorithm

Background and aims: Pulmonary tuberculosis is the biggest cause of death from infectiousdiseases and is one of the ten main causes of death in the world. Reducing the delay in definitivediagnosis is very important in reducing the transmission process of the disease and minimizingthe rate of its reproduction. Therefore, creating a diagnostic aid system using artificial intelligencetechniques to screen for tuberculosis can help in the early diagnosis of this disease. The presentstudy was conducted with the aim of investigating the Naive Bayes algorithm for intelligent diagnosisof this disease.Method: In order to conduct this study, the research population of patients with tuberculosissymptoms and the research sample is the recorded data of ۴۶۲ patients with early symptoms oftuberculosis retrospectively from May to the end of March ۲۰۱۹ in the database of the TabrizTuberculosis and Pulmonary Diseases Research Center. The information of the samples withconfirmed diagnosis was checked in two classes of pulmonary tuberculosis and normal. NaiveBayes algorithm has been used to screen for pulmonary tuberculosis using the general and primarysymptoms of patients using the Python programming language.Results: In the implementation of the Naive Bayes algorithm for the diagnosis of pulmonarytuberculosis, the sensitivity, accuracy and specificity of the result were ۹۴.۴۹%, ۹۵.۶۳% and۹۸.۵۶%, respectively, and the Area Under the ROC Curve (AUC) was calculated as ۹۸.۹۲%.Conclusion: The performance of the simple Bayes model for the diagnosis of pulmonary tuberculosishas acceptable accuracy. Since tuberculosis disease is relatively common in our country,timely diagnosis of this disease plays a significant role in its treatment and management, especiallyin remote areas with limited access to laboratory resources and lack of specialists. As aresult, the rapid and accurate development of new diagnostic tools and techniques in tuberculosisis essential.

کلیدواژه ها:

Pulmonary Tuberculosis ، Naive Bayes Algorithm ، Artificial Intelligence