A telemedicine tool framework for lung sounds classification using ensemble classifier algorithms

Abstract

Abstract Telemedicine is one of the medical services related to information exchange tools (eHealth). In recent years, the monitoring and classification of acoustic signals of respiratory-related disease is a significant characteristic in the pulmonary analysis. Lung sounds produce appropriate evidence related to pulmonary disorders, and to assess subjects pulmonary situations. However, this traditional method suffers from restrictions, such as if the doctor isn't very much practiced, this may lead to an incorrect analysis. The objective of this research work is to build a telemedicine framework to predict respiratory pathology using lung sound examination. In this paper, the three approaches has been compared to machine learning for the detection of lung sounds. The proposed telemedicine framework trained through Bagging and Boosting classifiers (Improved Random Forest, AdaBoost, Gradient Boosting algorithm) with an extracted set of handcrafted features. The experimental results demonstrated that the performance of Improved Random Forest was higher than Gradient Boosting and AdaBoost classifiers. The overall classification accuracy for the Improved Random Forest algorithm has 99.04%. The telemedicine framework was implemented with the Improved Random Forest algorithm. The telemedicine framework has achieved phenomenal performance in recognizing respiratory pathology.