Indirect Detection of Swine Influenza Activity in Porcine Blood Using Raman Spectroscopy and Machine Learning.

Abstract

Over the past decade, several swine influenza variants, including H1N1 and H1N2, have been periodically detected in swine. Raman spectroscopy (RS) offers a non-destructive, label-free, and rapid method for detecting pathogens by analyzing molecular vibrations to capture biochemical changes in samples. In this study, we examined blood serum from swine under different conditions: healthy, unvaccinated, or vaccinated against porcine reproductive and respiratory syndrome, and vaccinated swine infected with H1N1 and H1N2 variants of swine influenza. Our findings demonstrate that RS, when combined with machine learning algorithms such as partial least squares discriminant analysis and eXtreme gradient boosting discriminant analysis, can achieve accuracy rates of up to 97.8% in identifying the infection status and specific variant within porcine blood serum. This research highlights RS as a useful, novel tool for the detection of influenza variants in swine, significantly enhancing surveillance efforts by identifying animal health threats.