Raman Spectral Features of Human Serum at Different Excitation for Identification of Chronic Heart Failure

This work analyzed the Raman spectral features of blood serum stimulated by 532 and 785 nm laser excitation sources to identify chronic heart failure (CHF). Binary models were constructed using the method of projections onto latent structures PLS-DA to classify 30 blood spectra from 10 patients with CHF and 30 blood spectra from 10 control patients. It was found that the ROC AUC of the model based on the analysis of Raman spectra stimulated by 532 nm radiation is 0.761, and by 785 nm radiation - 0.847. It was found that the Raman characteristics stimulated by the 785 and 532 nm wavelengths are not superior to each other in terms of information content, because the improvement in ROC AUC when analyzing spectra stimulated by 785 nm radiation is not statistically significant (p = 0.14).