Establishing Healthy Breath Baselines With Tin Oxide Sensors: Fundamental Building Blocks for Noninvasive Health Monitoring
Shivaum Heranjal,Mariana Maciel,Sai Nishith Reddy Kamalapally,Ishan Ramrakhiani,Eray Schulz,Sha Cao,Xiaowen Liu,Ryan F Relich,Ronald Wek,Mark Woollam,Mangilal Agarwal
DOI: https://doi.org/10.1093/milmed/usae078
2024-08-19
Abstract:Introduction: Volatile organic compounds (VOCs) in breath serve as a source of biomarkers for medical conditions relevant to warfighter health including Corona Virus Disease and other potential biological threats. Electronic noses are integrated arrays of gas sensors that are cost-effective and miniaturized devices that rapidly respond to VOCs in exhaled breath. The current study seeks to qualify healthy breath baselines of exhaled VOC profiles through analysis using a commercialized array of metal oxide (MOX) sensors. Materials and methods: Subjects were recruited/consented through word of mouth and using posters. For each sample, breath was analyzed using an array of MOX sensors with parameters that were previously established. Data were also collected using a lifestyle questionnaire and from a blood test to assess markers of general health. Sensor data were processed using a feature extraction algorithm, which were analyzed through statistical approaches to identify correlations with confounding factors. Reproducibility was also assessed through relative standard deviation values of sensor features within a single subject and between different volunteers. Results: A total of 164 breath samples were collected from different individuals, and 10 of these volunteers provided an additional 9 samples over 6 months for the longitudinal study. First, data from different subjects were analyzed, and the trends of the 17 extracted features were elucidated. This revealed not only a high degree of correlation between sensors within the array but also between some of the features extracted within a single sensor. This helped guide the removal of multicollinear features for multivariate statistical analyses. No correlations were identified between sensor features and confounding factors of interest (age, body mass index, smoking, and sex) after P-value adjustment, indicating that these variables have an insignificant impact on the observed sensor signal. Finally, the longitudinal replicates were analyzed, and reproducibility assessment showed that the variability between subjects was significantly higher than within replicates of a single volunteer (P-value = .002). Multivariate analyses within the longitudinal data displayed that subjects could not be distinguished from one another, indicating that there may be a universal healthy breath baseline that is not specific to particular individuals. Conclusions: The current study sought to qualify healthy baselines of VOCs in exhaled breath using a MOX sensor array that can be leveraged in the future to detect medical conditions relevant to warfighter health. For example, the results of the study will be useful, as the healthy breath VOC data from the sensor array can be cross-referenced in future studies aiming to use the device to distinguish disease states. Ultimately, the sensors may be integrated into a portable breathalyzer or current military gear to increase warfighter readiness through rapid and noninvasive health monitoring.