Analysis of exhaled human breath for diagnosis of Tuberculosis disease from human breath

Tuberculosis (TB) is one of the leading causes of death worldwide and remains a global health challenge necessitating development of innovative diagnostic approaches for early and accurate detection of the disease. Current established diagnostic techniques are invasive, time-intensive, and contribute to delays in diagnosis, thereby exacerbating disease progression in patients and facilitating community transmission. To address these limitations, this study investigated human breath samples to identify volatile organic compounds (VOCs) associated with active TB as potential non-invasive biomarkers, with the ultimate goal of developing a non-invasive breathalyser diagnostic device. VOCs preconcentrated using solid phase microextraction (SPME) were analysed using gas chromatography-mass spectrometry (GC-MS), supported by AMDIS and OpenChrom software for compound identification. The findings revealed the presence of previously unreported VOCs associated with breath samples from patients with active TB and multidrug-resistant TB (MDR-TB), which were absent in control participants without TB symptoms. Furthermore, the results suggest the feasibility of differentiating MDR-TB from active TB based on breath VOC profiles, marking a novel observation. Machine learning models including Support Vector Machine (SVM), Random Forest (RF), Decision Trees (DT), and K-Nearest Neighbours (KNN) were evaluated for classifying breath samples according to their study groups, with the SVM achieving the highest performance. These findings underpin the development of a non-invasive breathalyser diagnostic device for TB, with the potential to enhance disease management and reduce transmission.

Atlang Gild Mpolokang is a PhD student in Physics at the Botswana International University of Science and Technology (BIUST), where he also works as a Teaching Assistant. His research focuses on application of advanced carbon-based nanomaterials for diagnosis of tuberculosis from exhaled human breath, aimed at developing a non-invasive diagnostics device for tuberculosis disease. He is the lead and co-author on peer-reviewed publications in Scientific Reports, including work on novel VOCs from exhaled breath of active TB patients (February 2025) and Determination of lung cancer exhaled biomarkers using machine learning – a new analysis framework (July 2025