[Quality assessment of the traditional Chinese medicine gentian by chemical pattern recognition]

Methanol extracts of 38 samples of the traditional Chinese medicine gentian of different origin were prepared. Data reflecting their chemical constituents as a whole were obtained by HPLC. The Chinese Pharmacopoeia (1990) specified four species of the getiana as authentic. Some samples from the authentic and non-authentic gentian were used as the training set of pattern recognition. Chemical features of the gentian samples were selected from among the matching data of HPLC by the SIMCA program. Finally, the selected chemical features were displayed as stars in a semicircular polar coordinate system, resulting in a constellation graph. Quality assessment of the traditional Chinese medicine gentian was accomplished by recognizing the position and path of the stars in the constellation graph. The clusters of the gentian samples in the constellation graph obtained from the HPLC features were consistent with and complement to their plant taxonomy. The vivid graph demonstrated that pattern recognition via chemical constellation promises well a reliable method for assessing traditional Chinese medicines.