ABSTRACT
The development of fast, cheap and reliable methods to determine seroconversion against infectious agents is of great practical importance. In the context of the COVID-19 pandemic, an important issue is to study the rate of formation of the immune layer in the population of different regions, as well as the study of the formation of post-vaccination immunity in individuals after vaccination. Currently, the main method for this kind of research is enzyme immunoassay (ELISA, enzyme-linked immunosorbent assay). This technique is sufficiently sensitive and specific, but it requires significant time and material costs. We investigated the applicability of attenuated total reflection (ATR) Fourier transform infrared (FTIR) spectroscopy associated with machine learning in blood plasma to detect seroconversion against SARS-CoV-2. The study included samples of 60 patients. Clear spectral differences in plasma samples from recovered COVID-19 patients and conditionally healthy donors were identified using multivariate and statistical analysis. The results showed that ATR-FTIR spectroscopy, combined with principal components analysis (PCA) and linear discriminant analysis (LDA) or artificial neural network (ANN), made it possible to efficiently identify specimens from recovered COVID-19 patients. We built classification models based on PCA associated with LDA and ANN. Our analysis led to 87% accuracy for PCA-LDA model and 91% accuracy for ANN, respectively. Based on this proof-of-concept study, we believe this method could offer a simple, label-free, cost-effective tool for detecting seroconversion against SARS-CoV-2. This approach could be used as an alternative to ELISA.