A Rapidly Method for the Discrimination of Aristolochic Acid and its Analogues Using SVM and PCA

Many traditional Chinese herb medicines containing aristolochic acid (AA) have been implicated in multiple cancer types, especially in upper tract urothelial carcinomas. The detection of AA and its analogues is of significant for the correct use of the drugs in the clinical Chinese medicine. In this paper, a nondestructive identification method based on the near-infrared spectroscopy (NIRS) technique, and the support vector machine (SVM) combined with the principal component analysis (PCA) is investigated to rapidly discriminate the AA and its analogues (denoted as PCA-SVM model). Firstly, PCA is developed to extract the effective wavelength variables according to the loading plots. The SVM model optimized by the grid search algorithm (Grid) is then applied to establish the qualitative analysis model. The experimental results demonstrate that the SVM model based on the Grid presents the excellent discrimination rate (100%) and minimum time comparing to the SVM optimized by the genetic algorithm and particle swarm optimization. Additionally, the variable number of the PCA model is validly reduced with a wavelength variable of 65. The PCA-SVM model based on the Grid is a suitable model to rapidly and efficiently discriminate AA and its analogues.