Method Study Of Vis-NIR-MIR Spectroscopy Applied To Analysis Of Albumin In Human Serum

Human serum albumin(ALB) is an important biochemical indicator of human health. Existing clinical determination methods require consuming of biochemical reagents and biochemical reactions, so that it is not convinient for the rapid detection of large crowd. Near-infrared spectrum(NIR) and fourier transform infrared spectrum attenuated total reflection(FTIR/ATR) technolodge, which are rapidly developing in recent years, has the advantage of chemical-free analysis of samples. There are existing works about feasibility of human serum albumin using the above methods. For human serum sample is complex object with multiple components, the effective chemometrics methods must be applied to extract information and mitigate noise disturbance. Therefore, an appropriate wavelength selection is an important and difficult technical aspect to improve the prediction effectiveness.In the paper, based on visible-near-infrared(Vis-NIR) and FTIR/ATR spectrum, the method used in chemical-free and rapid quantitative analysis of human serum albumin has been systematically studied. Three original wavelength optimization methods was proposed, namely equidistant combination partial least squares(EC-PLS), the absorbance combination partial least squares(AO-PLS) and correlation coefficient combination partial least squares(CCO-PLS). The above methods, which are choosed or combined, achieve excellent analysis results in the analysis of the relevant spectral analysis.For the Vis-NIR spectrum analysis of ALB,(1) based on EC-PLS method, the initial wavelength(I), number of wavelengths(N), number of wavelength gaps(G) of the optimized wavelength model were 1550 nm, 37,5 respectively. The corresponding root-mean-square errors(SEP) and correlation coefficients(RP) and relative SEP(RSEP) for prediction were 0.52 g L-1,0.995 and 1.32% respectively;(2) based on AO-PLS method, the number of wavelength for optimization model was 98, the corresponding SEP, RP, RSEP were 0.48 g L-1,0.996 and 1.21%;(3) based on AO-EC-PLS method, the optimized number of wavelength gaps and number of wavelengths were 3 and 28, the corresponding SEP, RP, RSEP were 0.46 g L-1,0.996 and 1.17%. The results show that the hyphenated AO-EC-PLS method achieved the best prediction effect and had the lowest model complexity.For the FTIR/ATR spectrum analysis of ALB,(1) based on EC-PLS method, the I, N and G of the optimized wavelength model were 2214 cm-1,186,4 respectively. the corresponding SEP, RP, RSEP were 1.13 g L-1,0.976 and 2.85% respectively;(2) based on AO-PLS method, the number of wavelength for optimization model was 557, the corresponding SEP, RP, RSEP were 1.10 g L-1,0.978 and 2.79%;(3) based on CCO-PLS method, the number of wavelength for optimization model was 338, the corresponding SEP, RP, RSEP were 1.10 g L-1,0.976 and 2.79%;(4) based on AO-EC-PLS method, the optimized number of wavelength gaps and number of wavelengths were 5 and 85, the corresponding SEP, RP, RSEP were 1.09 g L-1,0.979 and 2.76%;(5) based on CCO-EC-PLS method, the optimized number of wavelength gaps and number of wavelengths were 6 and 46, the corresponding SEP, RP, RSEP were 1.05 g L-1,0.978 and 2.65%. The results show that the hyphenated CCO-EC-PLS method achieved the best prediction effect and had the lowest model complexity.In conclusion, based on Vis-NIR and FTIR/ATR spectrum, the precise quantitative model for analysis of albulin in human serum has established via systematic study of optimization of analytical wavelength model. This method, which is chemical-free and rapid, provides valuable the basis of method for the analysis of albulin in human serum. In the paper, the established algorithm and platform of computer programming hopefully applied to the analysis of other objects.