拉曼光谱快速鉴别品牌药与仿制药的研究

Rapid Screening of the branded and generic drugs by Raman Spectroscopy

本文用拉曼光谱结合随机森林法很好的区分了相同药品不同厂家之间差异,为拉曼光谱用于快速区分仿制药和品牌药奠定基础。选用了1个品牌药厂家和6个仿制药厂家生产的卡托普利片作为实验研究对象,经过截取拉曼波长范围、光谱平滑、背景扣除、中心化等预处理,优化参数后,用随机森林法和主成分分析投影判别分别建立分析模型。计算结果表明:随机森林法所建立的分类模型可以区分卡托普利片的品牌药与仿制药厂家间的差异性,结果优于的主成分分析降维后不同主成分数下的投影判别分析效果。因此结合随机森林法能为拉曼光谱在药品快检中的高效、快速、无损分析提供有力保障。

In this paper, Raman spectroscopy in combination with Random forest method were used to distinguish the differences between the same drugs of different manufacturers efficiently. The captopril tablets were used as the experimental samples, and one of them was branded version, the others were generic drug manufacturers. Additionally, the pretreatment methods of spectra wavelength range cutting, smoothing, and baseline correcting, mean-centralizing and so on. After optimization of parameters, the prediction model was established by method of Principal component analysis （PCA） and Random forests （RF）. The results show that PCA cannot differentiate the branded version and six generic drug manufacturers, while the RF characterized the differences very well. So one can draw a conclusion that the application of RF can make a great contribution to Raman spectroscopy for the drug fast screening efficiently and accurately.