Extraction and identification of information from Mass Spectra of the breath of patients infected with SARS-CoV-2

After the pandemic of COVID-19, rapid COVID screening plays an essential role of control the spread. However, almost all the current screening methods are invasive and bring discomfort and potential harm to subjects, especially for vulnerable patients such as infants and the aged. Therefore, a non-invasive detection method based on volatile organic components (VOC) is researched. Using specific equipment proposed by NanoTech Analysis S.r.l. (NTA) and the collected volatile gas plastic bag from cooperated health care center, the Gas Chromatography Mass Spectrometry (GC-MS) of VOC are generated and further processed as fingerprint of each subject. Consequently, with 89 positive and 230 negative patients’ data, several machine learning algorithms (random forest, support vector machine, gradient boosting machine, etc.) are combined to acquire accuracy ranging from 80% to 90%, and recall ranging from 70% to 85%. The developed technology provides a novel concept for non-invasive rapid test screening for COVID-19 in various scenarios, although more positive data is required to tune the model.