Raman spectroscopy based discrimination of NS1 positive and negative dengue virus infected serum

Abstract

This study is intended to develop a multivariate statistical model for the prediction of nonstructural protein 1 (NS1) in dengue virus (DENV) infected blood serum in humans. The model has been developed on the basis of partial least squares regression using the Raman spectra of NS1 positive and NS1 negative samples. Human blood sera of 218 subjects is included in this study, of which 95 were NS1 positive and 123 were NS1 negative, which was confirmed with the enzyme linked immunosorbent assay method. For model development, 80 NS1 positive and 98 NS1 negative samples were used, while 40 DENV suspected samples were used for double blind testing of the model. This selection of samples was performed by the code in an automatic manner to avoid biasing. A laser at 785 nm was used as the excitation source to acquire Raman spectra of samples with an integration time of 15 s. The multivariate model yields coefficients of regression at corresponding Raman shifts. These coefficients represent changes in the molecular structures associated with NS1 positive and negative samples. The analysis of the regression coefficients which differentiate NS1 positive and NS1 negative groups shows an increasing trend for phosphatidylinositol, ceramide, and amide-III, and a decreasing trend for thiocyanate in the DENV infected serum. The R-squared value of the model was found to be 0.91, which is clinically acceptable. The blind testing of 40 suspected samples yields an accuracy, sensitivity, and specificity of about 100% each.