BACKGROUND/OBJECTIVES: Despite the prevalence of multiple sclerosis, there is currently no biomarker by which this disease can be reliably identified. Existing diagnostic methods are either expensive or have low specificity. Therefore, the search for a diagnostic method with high specificity and sensitivity, and at the same time not requiring complex sample processing or expensive equipment, is urgent.

METHODS

The article discusses the use of blood serum surface enhanced Raman spectroscopy in combination with machine learning analysis to separate persons with multiple sclerosis and healthy individuals. As a machine learning method for Raman spectra processing the projection on latent structures-discriminant analysis was used.

RESULTS

Using the above methods, we have obtained possibility to separate persons with multiple sclerosis and healthy ones with an average specificity of 0.96 and an average sensitivity of 0.89. The main Raman bands for discrimination against multiple sclerosis and healthy individuals are 632, 721-735, 1,048-1,076 cm. In general, the study of the spectral properties of blood serum using surface enhanced Raman spectroscopy is a promising method for diagnosing multiple sclerosis, however, further detailed studies in this area are required.