Research On Near Infaraed Spectroscopy Analysis Of Glycosylated Serum Protein

Blood glucose detection well can diagnose and prevent the occurrence of diabetes effectively, and delay the development of acute diabetes and chronic complications. The detection of glycosylated serum protein can reflect the level of blood glucose control in the near future. In this paper, the specificity of glycosylated serum protein is studied and the positions of characteristic absorption peak are determined by two-dimensional correlation spectroscopy and density functional theory. Combining with chemo metric methods, the quantitative correction model of glycosylated serum protein concentration is established, and prediction ability of the model is tested with the verification set. The main research works are as follows:(1)The spectrum of glucose, serum protein and glycosylated serum protein is analyzed and the absorption peaks are assigned by two-dimensional correlation spectroscopy analysis and density functional theory calculation from theory and experiment respectively. And the absorption peak caused by the characteristic structure N-H group of glycosylated serum protein is in the vicinity of 7053 cm-1. In the calculation of density functional theory, the three fundamental absorption peaks of the glycosylated serum protein compared to glucose, serum protein with significant changes are 2953.91 cm-1, 3101.44 cm-1 and 3528.14 cm-1, its double frequency are consistent with the wave number 5923 cm-1, 6332 cm-1 and 7053 cm-1 which are obtained by the 2D correlation spectroscopy analysis.(2) In determining the glycosylated serum protein specificity and the characteristic absorption peak positions, the genetic algorithm and interval partial least squares (GA-iPLS) is adopt to establish the quantitative calibration model of glycosylated serum protein, its determination coefficient is 0.9763, and the RMSEC is 0.007. The optimal wavelength band of the genetic algorithm includes the absorption peaks of two dimensional correlation spectroscopy analysis and density functional theory calculation, it shows that the preferred band can be used for the establishment of glycosylated serum protein concentration model well. The prediction ability of the model is tested by using the validation set. The determination coefficient of the prediction model is 0.9705 and the RMSEP is 0.0087. The results show that the concentration model of near infrared spectroscopy can be used in the clinical detection of glycosylated serum protein.