Analysis on Processing Method of Spectrum-Effect Relationship for Material Basis of Traditional Chinese Medicine

This paper summarizes the existing screening methods of active substances in traditional Chinese medicine (TCM). Most of the screening methods involve a great deal of data, which can be classified as small sample data with high-dimension in most cases. Data mining is considered as a better method for material basis screening of Chinese herbal compound ingredients after analyzing data processing methods on spectrum-effect relationship. After the initial classification of data by data mining means, the errors curves of spectrum-effect could be divided into two categories. One is the ideal case where the errors are normally distributed and the other is the non-ideal case where the errors are not normally distributed. For different types of error curves, different models can be selected to obtain the dose-utility relationship. At last, we analyzed the possible research ideas in the treatment of the dose-effect relationship.