chest diseases diagnosis based on machine learning algorithms

Several diseases threaten human health by affecting longevity and its well-being in many ways. Among them, chest diseases as Tuberculosis (TB), Chronic Obstructive Pulmonary Disease (COPD), Pneumonia, Asthma, and Lung cancer are considered as serious health complications and one major cause of death in both developing and developed countries. Doctors confirm that the earlier a disease is diagnosed, the higher is the patient cure probability. In this context, expert systems and different artificial intelligence techniques have been successfully used to solve different problems in various domains including medical diagnosis. In this thesis, we use the Support Vector Machines (SVM), the K-Nearest Neighbors (K-NN) and the Naive Bayes classification (NB) methods to diagnose chest diseases and for the first time, we examine the performance of the Adaptive Support Vector Machine (ASVM) method for chest disease diagnosis. This involves improving the SVM by finding its most appropriate Bias term value. These approaches are evaluated using an experimental dataset from Diyarbakir chest diseases hospital and comparing them with the Neural Network method used in previous studies. The experimental results showed the efficiency of these methods, especially ASVM, which could achieve promising results and confirmed that it can be efficiently used in chest diseases.