FTIR结合SVR对三七总多糖含量快速预测

对中药进行快速质量控制, 从整体层面反映中药的安全性与有效性具有重要意义。通过硫酸-苯酚显色反应测定三七总多糖含量, 傅里叶变换红外光谱(FTIR)结合支持向量机回归(SVR)建立三七总多糖含量预测模型, 以期为三七提供快速准确的质控方法。采集云南省12个产地60个三七样品的红外光谱, 紫外分光光度法(UV-Vis)检测样品中总多糖含量。红外光谱经过二阶导数(2D)、正交信号校正(OSC)、小波变换(WT)和变量投影重要性(VIP)筛选等数据优化处理。 SPXY算法将所有样本按2∶1的比例划分为训练集与预测集。训练集数据用于建立SVR预测模型, 网格式搜索、遗传算法(GA)和粒子群优化算法(PSO)对SVR预测模型进行参数优化, 预测集进一步对SVR模型的预测能力进行验证。结果显示: (1)葡萄糖标准品与三七总多糖在490 nm处存在最大共有吸收峰, 490 nm可作为三七总多糖检测的定量波长； (2)文山丘北、曲靖师宗及红河蒙自等产地的三七总多糖含量较高, 平均含量在25 mg·g-1以上； (3)分析3种参数优化模型的校正均方根误差(RMSEE)与预测均方根误差(RMSEP), 与PSO优化模型相比, 网格式搜索优化模型欠学习, GA优化模型过学习； (4)PSO-SVR模型对预测集数据预测效果最好, RMSEP=3.120 6, R2pre=83.13%, 预测值与紫外检测值接近。表明FTIR结合PSO-SVR模型能够对三七中总多糖含量进行快速准确的预测, 为保证三七稳定、安全与有效用药提供数据。

Abstract

The multi-component synergy is one of the important pathways for the pharmacological effects of traditional Chinese medicine (TCM) due to the complicated chemical compositions. Therefore, it is necessary to control and reflect the quality of TCM comprehensively in order to ensure its efficacy and safety. In Chinese Pharmacopoeia, the contents of three saponins were selected as indicators to ensure the quality of P. notoginseng. However, a single type indicator was limited to evaluate the quality of P. notoginseng comprehensively. In this study, the total polysaccharides content of P. notoginseng was determined by using ultraviolet-visible (UV-Vis) spectroscopy and phenol sulfuric acid reaction, and a prediction model of total polysaccharides content was established to provide some basic researches for rapid and comprehensive quality assessment of P. notoginseng based on Fourier transform infrared (FTIR) spectroscopy combined with support vector regression (SVR). In addition, a total of 60 FTIR spectra of P. notoginseng originated from 12 regions were collected. The absorbance of UV-Vis spectra at 490 nm which was contributed by polysaccharide extraction solution was recorded, and the content of total polysaccharides was calculated based on standard linear equation of glucose. Moreover, optimization procedures of spectra data were calculated by second derivative (2D), orthogonal signal correction (OSC), wavelet transform (WT), and variable importance for the projection (VIP). 2/3 of the 60 individuals were selected to develop the calibration set by using SPXY algorithm, and the rest samples were used as validation set. Calibration set data was used to establish the SVR model and grid search, genetic algorithm (GA) and particle swarm optimization algorithm (PSO) were used for screening optimal parameters which were utilized to verify the accuracy and reliability of the SVR model. Results showed that: (1) Maximum absorption peaks of glucose and total polysaccharides were both at 490 nm, and therefore the absorbance of UV-Vis spectra at 490 nm could be used for calculating the content of total polysaccharides. (2) The P. notoginseng from Qiubei, Shizong and Mengzi origins contained higher content of total polysaccharides (more than 25 mg·g-1) than other producing origins. (3) By analyzing the root mean square error of estimation (RMSEE) and the root mean square error of prediction (RMSEP) of optimization model, we found that the grid search model were under-fitting and the GA model were over-fitting compared with PSO model. (4) PSO model showed an excellent predictive effect with RMSEP and R2pre of 3.120 6 and 83.13% respectively, which indicated the predicted values were close to the detection values. The result indicated that FTIR combined with PSO-SVR could accurately predict the content of total polysaccharides, which could provide a research basis for the comprehensive quality control as well as ensure the stable, safe and effective medicinal use of P. notoginseng.

参考文献

[1] Jiang Y, David B, Tu P F, et al. Analytica Chimica Acta, 2010, 657(1): 9.

[2] Yan Y, Chai C Z, Wang D W, et al. Journal of Pharmaceutical and Biomedical Analysis, 2013, 80: 192.

[3] Cao G, Zhang C, Zhang Y, et al. Journal of Separation Science, 2011, 34(15): 1845.

[4] Uzayisenga R, Ayeka P A, Wang Y. Phytotherapy Research, 2014, 28(4): 510.

[5] Wang Z, Chen Y Y, Pan H J, et al. Chinese Herbal Medicines, 2015, 7(2): 179.

[6] Chinese Pharmacopoeia Commission(中华人民共和国药典委员会). Pharmacopoeia of the People’s Republic of China, Part One (中华人民共和国药典一部). Beijing: China Medical Science Press(北京: 中国医药科技出版社), 2015.

[7] Wang J R, Yau L F, Gao W N, et al. Journal of Agricultural and Food Chemistry, 2014, 62(36): 9024.

[8] Gao H, Wang F, Lien E J, et al. Pharmaceutical Research, 1996, 13(8): 1196.

[9] LI Shi-ying, XIE Yun-liang, YU Zhong-hui, et al(李世英, 谢云亮, 于忠慧, 等). Journal of Apoplexy and Nervous Diseases(中风与神经疾病杂志), 2015, 32(11): 996.

[10] HU Feng-xia, ZHU Shuang(胡凤霞, 朱双). Journal of Liaoning University of Traditional Chinese Medicine(辽宁中医药大学学报), 2015, 17(10): 32.

[11] WEI Wei, HUANG Yong-zhi, LIANG Liu-dan, et al(韦维, 黄永秩, 梁柳丹, 等). Tianjin Medical Journal(天津医药), 2014, 42(9): 893.

[12] Wu Y, Wang D. Journal of Natural Products, 2008, 71(2): 241.

[13] Li Y, Zhang J Y, Wang Y Z. Analytical and Bioanalytical Chemistry, 2018, 410(1): 91.

[14] LI Yun, ZHANG Ji, XU Fu-rong, et al(李运, 张霁, 徐福荣, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2017, 37(1): 70.

[15] LI Yun, ZHANG Ji, XU Fu-rong, et al(李运, 张霁, 徐福荣, 等). Chinese Traditional and Herbal Drugs(中草药), 2017, 48(22): 4764.

[16] Guo Y J, Deng G F, Xu X R, et al. Food & Function, 2012, 3(11): 1195.

[17] Mu L, Liu L, Ma D. Journal of Forestry Research, 2010, 21(1): 77.

[18] LI Jiao, WANG Ke, WANG Rui-po, et al(李姣, 王珂, 王瑞坡, 等). Food Science(食品科学), 2011, 32(8): 65.

[19] Cheng C, Liu J, Zhang C, et al. Applied Spectroscopy Reviews, 2010, 45(2): 148.

[20] Kyomugasho C, Christiaens S, Shpigelman A, et al. Food Chemistry, 2015, 176: 82.

[21] Luo Q, Yun Y, Fan W, et al. RSC Advances, 2015, 5(7): 5046.

[22] Kim D Y, Cho B K, Lee S H, et al. Sensors and Actuators B: Chemical, 2016, 230: 753.

[23] Guo T, Feng W H, Liu X Q, et al. Journal of Pharmaceutical and Biomedical Analysis, 2015, 123(1): 16.

[24] Chen Y, Xie M, Zhang H, et al. Food Chemistry, 2012, 135(1): 268.

[25] Zhuang H, Ni Y, Kokot S. Applied Spectroscopy, 2015, 69(4): 488.

[26] SUN Su-qin(孙素琴). Analysis of Traditional Chinese Medicine by Infrared Spectroscopy(中药红外光谱分析与鉴定). Beijing: Chemical Industry Press(北京: 化学工业出版社), 2010.

[27] Cortes C, Vapnik V. Machine Learning, 1995, 20(3): 273.

[28] Kalteh A M. Water Resources Management, 2015, 29(4): 1283.

[29] Liu M, Wang M, Wang J, et al. Sensors and Actuators B: Chemical, 2013, 177: 970.

[30] Shi J, Zhou X, Huang X, et al. Food Chemistry, 2013, 138(1): 192.

[31] Chang C C, Lin C J. ACM Transactions on Intelligent Systems and Technology (TIST), 2011, 2(3): 27.

[32] MATLAB Chinese BBS(MATLAB中文论坛). MATLAB 30 Case Analysis of Neural Network(MATLAB神经网络30个案例分析). Beijing: Beihang University Press(北京: 北京航空航天大学出版社), 2010.

[33] Galvao R K H, Araujo M C U, Jose G E, et al. Talanta, 2005, 67(4): 736.

[34] Li D, Jin Z, Zhou Q, et al. Journal of Molecular Structure, 2010, 974(1): 68.

[35] Shen F, Yang D, Ying Y, et al. Food and Bioprocess Technology, 2012, 5(2): 786.

[36] Wang G, Yin S. IEEE Transactions on Industrial Informatics, 2015, 11(2): 398.

[37] Haufe S, Dhne S, Nikulin V V. NeuroImage, 2014, 101: 583.

李运, 张霁, 刘飞, 徐福荣, 王元忠, 张金渝. FTIR结合SVR对三七总多糖含量快速预测[J]. 光谱学与光谱分析, 2018, 38(6): 1696. LI Yun, ZHANG Ji, LIU Fei, XU Fu-rong, WANG Yuan-zhong, ZHANG Jin-yu. Prediction of Total Polysaccharides Content in P. notoginseng Using FTIR Combined with SVR[J]. Spectroscopy and Spectral Analysis, 2018, 38(6): 1696.

FTIR结合SVR对三七总多糖含量快速预测

关于本站 Cookie 的使用提示

全站搜索

FTIR结合SVR对三七总多糖含量快速预测

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索