Machine learning assisted rapid approach for quantitative prediction of biochemical parameters of blood serum with FTIR spectroscopy
O.G. Chechekina,E.V. Tropina,L.I. Fatkhutdinova,M.V. Zyuzin,A.A. Bogdanov,Y. Ju,K.N. Boldyrev
DOI: https://doi.org/10.1016/j.saa.2024.125283
IF: 4.831
2024-10-19
Spectrochimica Acta Part A Molecular and Biomolecular Spectroscopy
Abstract:This study develops regression models for predicting blood biochemical data using Fourier-transform infrared spectroscopy (FTIR) analysis. Absorption at specific wavelengths of blood serum is revealed to have strong correlations with biochemical parameters, such as ALT, amylase, AST, protein, bilirubin, Gamma-GT, iron, calcium, uric acid, triglycerides, phosphatase and cholesterol, were shown. The results consistently demonstrate that Random Forest Regression outperforms other models, delivering impressive outcomes for the majority of the analyzed parameters. For some parameters we obtained a coefficient of determination of 0.95 and more (amylase, AST, iron, calcium, protein, uric acid and cholesterol), which makes this approach to be applicable in clinical diagnostics. These findings highlight the potential of FTIR analysis combined with regression models for precise assessment of blood biochemistry.
spectroscopy