Determination of Proteins in Human Serum by Near-Infrared Spectroscopy with Partial Least Square Analysis

Abstract

ABSTRACTNear-infrared spectroscopy with interval and synergy interval partial least squares were used to determine total proteins, albumin, and globulin in human serum. Quantitative models of the relationships between the near-infrared spectra and these proteins were established. The root mean square error of prediction (RMSEP) was used to evaluate the models. The results demonstrate that synergy interval partial least squares was superior to interval partial least squares. The optimum wavelength ranges using synergy interval partial least squares for total proteins, albumin, and globulin in serum were 1668–1772 and 1772–1874 nm; 812–1000 and 1636–1800 nm; and 1228–1364 and 1642–1772 nm, respectively. The RMSEP of these proteins were 0.6905, 0.6686, and 0.5355 g/L. Waveband optimization was necessary to enhance the accuracy. The synergy interval partial least square model for total proteins, albumin, and globulin in serum offered higher accuracy and overcomes shortcomings of interval partial least squares. Furthermore, the optimized spectrum region may allow the design of near-infrared spectrophotometers for specific applications.