Prediction of Chronic Respiratory Diseases Using Machine Learning Algorithms

This research attempts to perform classification on four groups of saliva samples using feature selection and machine learning algorithms. These groups primarily consist of COPD (Chronic Obstructive Pulmonary Disease), Asthma, Infected, and HC (Healthy Control). We aim to diagnose a particular patient with the type of respiratory disease with the help of machine learning models like Naïve Bayes Classifier, Decision Tree Classifier, XGboost Classifier, and Random Forest Classifier. In this work, we apply the previously mentioned machine learning methods to the exasens dataset. To further enhance our predictive model’s results, we have used feature selection methods- Intrinsic feature selection, Recursive Feature Elimination (Wrapper method), Kendall feature selection, and Anova feature selection. We have also used various data visualization methods for better comprehension of our models. Consequently, we have obtained the highest classification accuracy for our diverse classification models when utilized alongside Anova Feature Selection.