Quantitative and qualitative analysis of IR spectra for medical and ecological applications
Yury V. Kistenev1, Denis A. Vrazhnov1, Viktor V. Nikolaev1, Akim A. Tretyakov1, Georgy K. Raspopin1, Didar R. Makashev1, Alexey V. Borisov1
1LMIML Laboratory, Tomsk State University, Tomsk, Russia
Abstract
IR spectra of gas samples are often referred to as a fingerprint region, which contains numerous absorptions lines resulting from stretching and bending vibrations of various single bonds in molecules of interest. Qualitative and quantitative analysis of such spectra allows estimating gas sample chemical composition and concentration of its components. The major problem of in this field is that the composition of gas samples of natural origin is unknown. This problem is closely related to the field of so cold a “gray system” analysis.
The developed methods of gray system analysis are effective when analyzed spectrum contains only 2-3 components. The more general problem of presence of more than 3 unac-counted (latent) components requires more sophistical methods, including machine learning (ML), first of all, aimed on extracting new informative variables (features) of minimum quan-tity, which describe peculiarities in data analogously to initial variables.
Original chemometrics and ML methods suitable for a quite general gray system analysis will be presented including examples of their medical and ecological applications.
The work was conducted with the financial support of the Ministry of Science and Higher Education of Russia (Agreement No. 075-15-2024-557 dated 04/25/2024)
Speaker
Yury V. Kistenev
Tomsk State University
Russia
Discussion
Ask question
