【摘要】 由于传统的组分分离-测试方法效率低并难以发现组分间的协同关系,逐步回归等传统统计分析方法又不可靠.用因果关系发现算法发现了关键化学组分,针对组效关系数据多为高维小样本的特点,提出了一种新的关系发现算法,并用此算法成功辨识出一种中药材川芎的关键组分,证明了该方法的可行性.更多还原

【Abstract】 A novel approach,i.e.applying causal discovery algorithms to identify key chemical compositions from composition-effect datasets,was proposed.To cope with the characteristics of high-dimension,small size of samples of composition-effect datasets,a new algorithm for discovering causal relationships,STEPCARD(STEPwise Causal Relationship Discovery),was introduced.The application of the new algorithm on CHUANXIONG,a traditional Chinese medicine,resulted in finding out one of the most effective compositions of CHUANXIONG,hence testified the effectiveness of the proposed approach.更多还原