NIR Determination of Three Critical Quality Attributes in Alcohol Precipitation Process of Lonicerae Japonicae with Uncertainty Analysis

The aim of this study was to investigate the feasibility of near infrared (NIR) spectroscopy and multivariate calibration (MVC) techniques in monitoring and understanding of the alcohol precipitation process of water extract of Lonicerae Japonicae within the framework of ICH´s Quality by Design (QbD) and FDA´s process analytical technology (PAT). The contents of Chlorogenic acid, Luteoloside and soluble solid were identified as three critical quality attributes (CQAs) to be monitored. After a comparison of different spectra preprocessing methods, the raw NIR spectra were applied to multivariate modeling. By the interval PLS (iPLS) method, three characteristic wavebands, 8802~7605 cm^-1, 6001~6201 cm^-1 and 7605~7308 cm^-1, were selected for Chlorogenic acid, Luteoloside and soluble solid, respectively. PLS quantitative models based on these wavebands showed much improved performance. Moreover, prediction uncertainty was denoted and visualized by both the PLS confidence interval and simple interval calculation (SIC) interval. And a new index-Extent of Uncertainty (E_u) is proposed to assess the magnitude of uncertainty for CQAs with different dimensions in the same scale. The overall results provided the useful understanding of and deep insight into the alcohol precipitation process of Chinese herbal medicine (CHM).