中药指纹图谱相似度评价方法的比较  

Comparison of different methods for evaluating the similarity of the fingerprints of traditional Chinese medicines

作　　者：聂磊[1] 曹进[1] 罗国安[1] 王义明[1] 

机构地区：[1]清华大学药物研究所,北京100084

出　　处：《中成药》2005年第3期249-252,共4页Chinese Traditional Patent Medicine

摘　　要：目的本文以痛必定粉针(高效液相色谱指纹图谱为研究对象,对各种指纹图谱的相似度评价方法进行了比较。方法先通过理论分析,比较了各种相似度系数的计算方法。如距离系数法、夹角余弦法、相关系数法、Nei系数法、改进Nei系数法。然后采用数字模拟以及统计方法对各种方法的性能进行了综合评价。结果各种相似度评价方法都有各自的特点和不足,其中距离系数具有较好的综合性质。结论实际应用中,应针对不同的实际问题,采用适合的相似度评价方法,以便更好的用于中药质量控制。AIM: In this context, the high performance liquid chromatography (HPLC) of Tongbiding Injection was used as fingerprint examples of traditional Chinese medicine to compare the different methods for evaluating the similarity. METHODS: Computing methods for similarity assessments were first compared on the basis of theoretical analysis, and then comprehensive evaluations for performances of each method of similarity assessment were investigated via data simulation and statistics. RESULTS: The results manifested that each approach had its individual properties and limitations. Distance coefficient has better general properties as compared with other methods of assessing similarity. CONCLUSION: In practie, to estimate the similarity between the fingerprints of traditional Chinese medicines, the appropriate methods would be selected for the different problems, and therefore quality of traditional Chinese medicines would be better controlled.

关 键 词：中药指纹图谱 相似度评价 方法的比较 痛必定粉针 数理统计 

分 类 号：R927.2[医药卫生—药学]