Zhang Guang-ren, Wu Yun, Jin Rui-ting, Wan Qin, Wu Jin-feng, Xu Xiao-qian, Xiao Wei
Zhongguo Zhong Yao Za Zhi. 2015 Aug;40(16):3204-8.
Sanjie Zhentong capsules were scanned by using a near infrared spectra probe with different drug mass fraction and the spectral information of capsule shells and contents in it were obtained. Then partial least squares (PLS) models were developed for the prediction of mass fraction of Fritillariae Thunbergii Bulbus and Resine draconis in Sanjie Zhentong capsules. The correlation coefficient (r9c)) and root mean standard error( RMSEC) of 0.949 5, 0.958 2 and 4.742 4, 4.135 7. The models obtained correlation coefficient (r(v)) of 0.919 2, 0.936 7 and root mean square error (RMSECV) of 6.158 9, 5.037 3 respectively in the training set. The paired T test analysis of statistics showed that there were no significant difference between predictive values and measure values. The established models reflected a strong prediction performance and can meet the needs of the production.
采用近红外光谱探头对不同药物质量分数的散结镇痛胶囊进行扫描,获取胶囊壳及内容物的光谱信息。然后建立偏最小二乘法(PLS)模型,用于预测散结镇痛胶囊中浙贝母和龙血竭的质量分数。相关系数(r9c))分别为0.949 5、0.958 2,根均方误差(RMSEC)分别为4.742 4、4.135 7。所建模型在训练集中的相关系数(r(v))分别为0.919 2、0.936 7,根均方误差(RMSECV)分别为6.158 9、5.037 3。统计学配对T检验分析表明,预测值与测量值之间无显著差异。所建立的模型具有较强的预测性能,能够满足生产需要。