Yang Yuangui, Zhao Yanli, Zuo Zhitian, Wang Yuanzhong
Institute of Medicinal Plants, Yunnan Academy of Agricultural Sciences, Kunming 650200, China.
J AOAC Int. 2019 Mar 1;102(2):457-464. doi: 10.5740/jaoacint.18-0188. Epub 2018 Sep 18.
var. (PPY) is used in the clinical treatment of tumors, hemorrhages, and anthelmintic. The aim of this study is to determine total flavonoids of PPY in the Yunnan and Guizhou Provinces, China. In this study, total flavonoids were determined by UV spectrophotometry at first. Then, Fourier transform mid-infrared (FT-IR) based on various pretreatments include standard normal variate (SNV), first derivative (FD), second derivative (SD), Savitzky-Golay (SG), and orthogonal signal correction (OSC) were investigated. In addition, several relevant variables were screened by competitive adaptive reweighted sampling (CARS). The content of total flavonoids and selected variables of FT-IR were used to establish a partial least squares regression for PPY in different regions. The results indicated that CARS was an effective method for decreasing the variable of the database and improving the prediction of the model. FT-IR with pretreatment SNV + OSC + FD + SG had the best performance, with R2 > 0.9 and residual predictive deviation = 3.3515, which could be used for the predictive model of total flavonoids. Those results would provide a fast and robust strategy for the determination of total flavonoids of PPY in different geographical origin. Various pretreatments, including SNV, FD, SD, SG, and OSC, were compared; several relevant variables were selected by CARS; and the content of total flavonoids and selected variable were used to establish a partial least squares regression for PPY in different regions.
变种（PPY）用于肿瘤、出血和驱虫的临床治疗。本研究的目的是测定中国云南和贵州地区PPY中的总黄酮含量。在本研究中，首先采用紫外分光光度法测定总黄酮含量。然后，研究了基于各种预处理方法的傅里叶变换中红外光谱（FT-IR），这些预处理方法包括标准正态变量变换（SNV）、一阶导数（FD）、二阶导数（SD）、Savitzky-Golay平滑滤波（SG）和正交信号校正（OSC）。此外，通过竞争性自适应重加权采样（CARS）筛选了几个相关变量。利用总黄酮含量和FT-IR选择的变量建立了不同地区PPY的偏最小二乘回归模型。结果表明，CARS是一种有效减少数据库变量并提高模型预测能力的方法。采用预处理方法SNV + OSC + FD + SG的FT-IR性能最佳，R2 > 0.9，剩余预测偏差 = 3.3515，可用于总黄酮的预测模型。这些结果将为测定不同产地PPY中总黄酮含量提供一种快速且可靠的策略。比较了包括SNV、FD、SD、SG和OSC在内的各种预处理方法；通过CARS选择了几个相关变量；并利用总黄酮含量和选择的变量建立了不同地区PPY的偏最小二乘回归模型。
