Abstract:

 A new type identification method, dual  index grade sequence pattern recognition method, was applied for identification and quality evaluation of the traditional Chinese compound formula Shiquan Dabu Pills produced by different companies. This approach possessed the  advantages of both supervised and unsupervised pattern recognition methods, which could simultaneously cluster and classify the complex biological  samples without prior knowledge. In the process, the most similar sample sequence of each sample was determined depending on common peak ratio index of samples and normal distribution. Similar samples should have ‘most similar sample sequences’, by which the samples could be  efficiently  recognized. The identification and quality evaluation of components of 22 Shiquan Dabu Pill samples produced by  three  companies extracted with chloroform were  carried out  by  this  novel approach and their infrared fingerprint spectra. The correct recognition rate of the samples was 100％, while it  was only 72.7％ by a famous hierarchical clustering method, an unweighted pair-group method using arithmetic averages (UPGMA). The results showed that the  dual index grade sequence pattern recognition method together with infrared fingerprint spectra could accurately and simply  finish the identification and quality estimation of the traditional Chinese compound formula, without any prior knowledge related to the analyzed samples.

Key words:  fingerprint spectra, clustering analysis, pattern recognition, Shiquan Dabu Pills, identification