基于最大似然基的中医药图谱有效性判定模型

本文提出了一种基于最大似然估计理论建立的中药指纹图谱有效判定模型,用于改进目前方法中存在的检测算法对主成分析的不足,注重对药品的定性分析而对中药的定量分析不足的问题。在本方法中用最大似然方法提取样本特征,用主成分评价和数字差分段评价的方法来解决目前相似性评价方法中存在的不能反映主成分差异这一不足之处,在解决检测数据漂移的问题上,采用基于统计学原理的漂移量纲分析,提高了漂移误差检测的速度。通过对丹参的色谱数据的检测试验可以看出,新方法能够提取检测出各成分的特征峰、含量比值,各样本中主成分相似性,从试验数据上看取得较好的效果,对今后对中药的主成分研究提供了检测方法。

Chinese Medicine Fingerprint Effectiveness Detection Model Based on Maximum Likelihood Estimation

This paper proposes a Chinese medicine fingerprint effectiveness detection model based on maximum likelihood estimation to improve the current method of detection algorithms existing in the main component and to solve the problem that quantitative analysis to Chinese medicine is insufficient.In our method maximum likelihood estimation is used to extract sample characteristic.Principal component evaluation and digital differential sub-evaluation is to solve the the problem of current evaluation method of similarity which can＇t reflect the difference among the principal component.It also raise the rapid of drift error detection based on drift dimensional analysis in statistical principle.The new method can extract and detect the characteristic peak, content ratio and the similarity of main composition in the sample. It is a new detection method for principal components of traditional Chinese medicine in future.