Near-infrared spectroscopy has rich structure and composition information and is often used to measure hydrogen-containing organic substances'physical and chemical parameters. In recent years, it has been widely used in the quantitative analysis of complex solutions. However, the near-infrared spectroscopy analysis of complex solutions such as human blood, noise interference and redundant variables caused by strong background information seriously affect the spectral measurement and analysis of the sample itself, and affect the efficiency and accuracy of the analysis. Therefore, eliminating the interference of background noise to improve the accuracy of analysis has attracted great attention. In recent years, scholars at home and abroad have proposed many related methods based on chemometric methods. This article focuses on spectral preprocessing, variable optimization and modeling analysis. On the one hand, starting from traditional chemometric methods, we summarize and analyze the application and respective characteristics of these methods in the near-infrared spectroscopy quantitative analysis of complex solutions such as human blood, and provide a reference for research on improving the accuracy of quantitative spectral analysis.