Component Analysis of Raman Spectra of Blood Serum Using Multivariate Curve Resolution Method

This paper is devoted to the study of Raman spectroscopy of blood serum for the diagnosis of chronic heart failure (CHF). The aim of the research is to develop an effective approach that combines the advantages of Raman spectroscopy with intelligent data processing methods to create an automated diagnostic system. To extract informative features from the spectral data of blood serum, the method of multivariate curve resolution was used. Such machine learning methods as logistic regression, support vector machine, random forest and gradient boosting were used to classify the spectra. To improve the quality of the model, the stacking method was used, which combines the results of the basic models using logistic regression as a metaclassifier. The results showed that the proposed approach is highly effective in identifying patients with CHF, and also demonstrated significant improvements when adding clinical data.