基于谱效关系的赤芍质量标志物筛选及BP人工神经网络抗凝血药效预测模型的构建OA北大核心CSTPCD

Objective:To screen the anticoagulant quality markers of Paeoniae Radix Rubra based on spectral effect correlation a-nalysis and construct the BP artificial neural network model to predict the anticoagulant efficacy of Paeoniae Radix Rubra from different sources.Methods:The fingerprint of Paeoniae Radix Rubra was established by HPLC,and the anticoagulant efficacy were evaluated by using in vitro anticoagulant activity assay with the four coagulation items as the indexes.The anticoagulant quality markers of Paeoniae Radix Rubra were screened by gray correlation analysis,and the anticoagulant efficacy evaluation model of Paeoniae Radix Rubra was constructed based on BP artificial neural network.Results:A total of 22 common peaks were confirmed in the fingerprint of Paeoniae Radix Rubra,9 components of them were identified.Seven candidate quality markers for anticoagulation of Paeoniae Radix Rubra were screened by spectral effect correlation analysis,including gallic acid,catechin,ethyl gallate,paeoniflorin,benzoic acid,1,2,3,4,6-O-pentagalloylglucose,benzoylpaeoniflorin.The maximum relative error between the measured value and the predicted value of the BP arti-ficial neural network model was 10.02%,the minimum relative error was 0.1%,the correlation coefficients between the two groups of data were more than 0.85.Conclusion:In this study,the candidate quality markers of anticoagulant efficacy of Paeoniae Radix Rubra are identified,which can provide a reference for the quality evaluation of Paeoniae Radix Rubra;The constructed BP artificial neural net-work model can be used for the evaluation of anticoagulant efficacy of Paeoniae Radix Rubra.