OBJECTIVE: To establish signature pattern recognition model of cold-hot nature of herbal medicine. METHODS: High performance capillary electrophoresis fingerprints of 60 kinds of herbal medicine (30 kinds of cold, 30 of hot) were established, features of wavelength were screened, 6 analysis methods such as linear discriminant analysis (LDA), logistic discriminant analysis (Logistic-DA), principal component and linear discriminant analysis (PCA-LDA), partial least-squares discriminant analysis (PLS-DA), random forest (RF) and support vector machine (SVM) were used to establish and evaluate recognition model of cold-hot nature after data processing. RESULTS: SVM was proved to be a suitable means of recognition model of herbal medicine cold-hot nature based on data of HPCE fingerprints. Characteristic parameters of nature could be screened according to theoretical spectra signature of nature model, the characteristic regions of components of herbs with cold-heat nature could be identified in the HPCE fingerprint. The characteristic parameters of cold-hot nature were the identifying coefficient for specific retention time of the theoretical spectra of recognition model, identification coefficients greater than zero were for the cold marker, while that less than zero for the hot marker. CONCLUSION: The results imply that HPCE is a feasible and effective means for identification of cold-hot nature of Traditional Chinese medicine.
OBJECTIVE: To establish signature pattern recognition model of cold-hot nature of herbal medicine. METHODS: High performance capillary electrophoresis fingerprints of 60 kinds of herbal medicine (30 kinds of cold, 30 of hot) were established, features of wavelength were screened, 6 analysis methods such as linear discriminant analysis (LDA), logistic discriminant analysis (Logistic-DA), principal component and linear discriminant analysis (PCA-LDA), partial least-squares discriminant analysis (PLS-DA), random forest (RF) and support vector machine (SVM) were used to establish and evaluate recognition model of cold-hot nature after data processing. RESULTS: SVM was proved to be a suitable means of recognition model of herbal medicine cold-hot nature based on data of HPCE fingerprints. Characteristic parameters of nature could be screened according to theoretical spectra signature of nature model, the characteristic regions of components of herbs with cold-heat nature could be identified in the HPCE fingerprint. The characteristic parameters of cold-hot nature were the identifying coefficient for specific retention time of the theoretical spectra of recognition model, identification coefficients greater than zero were for the cold marker, while that less than zero for the hot marker. CONCLUSION: The results imply that HPCE is a feasible and effective means for identification of cold-hot nature of Traditional Chinese medicine.