Machine learning analysis of SERS fingerprinting for the rapid determination of Mycobacterium tuberculosis infection and drug resistance

• DOI: 10.1016/j.csbj.2022.09.031
• Publication Date: 2022-09-26T02:33:41Z
• Authors: Liang Wang; Xue-Di Zhang; Jia-Wei Tang; Zhang-Wen Ma; Muhammad Usman; Qing-Hua Liu; Chang-Yu Wu; Fen Li; Zuo-Bin Zhu; Bing Gu
• Keywords: Point of care

Over the past decades, conventional methods and molecular assays have been developed for detection of tuberculosis (TB). However, these techniques suffer limitations in identification Mycobacterium (Mtb), such as long turnaround time low sensitivity, etc., not even mentioning difficulty discriminating antibiotics-resistant Mtb strains that cause great challenges TB treatment prevention. Thus, with easy implementation rapid diagnosis infection are high demand routine diagnosis. Due to label-free, low-cost non-invasive features, surface enhanced Raman spectroscopy (SERS) has extensively investigated its potential bacterial pathogen identification. at current stage, few studies recruited handheld spectrometer discriminate sputum samples or without Mtb, separate pulmonary from extra-pulmonary strains, profile different antibiotic resistance characteristics. In this study, we a set supervised machine learning algorithms dissect SERS spectra generated via focus on deep algorithms, through which were successfully differentiated (5-fold cross-validation accuracy = 94.32%). Meanwhile, isolated effectively separated 99.86%). Moreover, drug-resistant profiles also competently distinguished 99.59%). Taken together, concluded that, assistance application point-of-care infections future.