The development and evaluation of an MOS-based electronic nose for the accurate discrimination of chronic obstructive pulmonary disease using breath analysis

This research paper presents a novel approach for discriminating patients with chronic obstructive pulmonary disease (COPD) from smokers and healthy controls using a self-made electronic nose device. This study aims to develop a portable and user-friendly system that accurately identifies patients with COPD based on volatile organic compound (VOC) profiles present in their breath. Breath samples were collected from 25 patients with COPD, 32 smokers, and 36 healthy controls. The MOS-based electronic nose device incorporated a sensor array consisting of TGS 2600, TGS 2610, TGS 2620, TGS 822, and TGS 826 sensors. Advanced signal processing techniques, including independent component analysis (ICA), were employed to analyze the breath samples and extract relevant features. Three classification models, namely the Support Vector Machine (SVM), Naive Bayes, and Decision Tree, were utilized to discriminate between patients with COPD, smokers, and healthy controls based on the extracted features. The results demonstrate the efficacy of the self-made MOS-based electronic nose device in accurately discriminating between patients with COPD, smokers, and healthy controls. The SVM model achieved a remarkable accuracy of 85.25% and an area under the curve (AUC) of 87%. This study highlights the potential of breath analysis to be used as a non-invasive and cost-effective approach for the diagnosis and differentiation of COPD. These findings provide a solid foundation for further research and the development of non-invasive breath analysis techniques in the field of respiratory disease diagnosis and monitoring.