Fast identification of influenza using label-free SERS combined with machine learning algorithms via clinical nasal swab samples

Abstract

Influenza virus outbreaks, which have become more frequent in recent years, have attracted global attention. Reverse transcription-polymerase chain reaction (RT-PCR) and enzyme-linked immunosorbent assay (ELISA), as the “gold standard” methods for virus detection, are not suitable for rapid diagnosis of the virus because of their long reaction time and sample preparation time. Therefore, a new method for influenza virus detection that is rapid, accurate and portable is needed. In this work, a label-free technology based on surface enhanced Raman spectroscopy was developed to directly analyse nasal swab samples in order to explore the molecular differences between influenza patients and normal people. Following that, machine learning algorithms based on Principal Component Analysis combined with Linear Discriminant Analysis (PCA-LDA) and Support Vector Machines (SVM) were used to extract and model the molecular features of nasal fluid to differentiate between influenza patients and normal people with an accuracy of 76.5%. With only 10 μL of sample and 5 seconds of testing time per sample, this label-free SERS combined with machine learning would provide a rapid and portable testing platform for influenza virus detection.