Comparison of the clinical effect features of Han-Ku-Gan and Wen-Xin-Gan based on the efficacy of promoting blood circulation and removing blood stasis

Objective:The application of Chinese materia medica(CMM)in clinical diseases is the embodiment and continuation of the property theory of CMM(PTCMM).However,due to a lack of precise quantitative description methods,it is difficult to systematically analyze the property of CMM(PCMM)and clinical effect features at the micro molecular level.Methods:The therapeutic drugs and targets were obtained from the Drugbank database.The molecular descriptors of these drugs were calculated based on Dragon software.Drug-effect relationships that integrated the molecular descriptors and effect descriptors were plotted as grayscale images.These images were used to train the Le Net-5 model and the Alex Net model.The best-performing model was used to predict the effect features of the CMM compounds.Finally,the effect features of the PCMM combinations were calculated based on the support vector machine recursive feature elimination algorithm.Results:The Alex Net model showed a superior prediction performance.The results showed that its accuracy,precision,sensitivity,F-measure,and Matthews correlation coefficient on the training set were 0.940,0.936,0.945,0.940,and 0.880,respectively,and those of the test set were 0.909,0.901,0.920,0.910,and 0.819,respectively.A total of 399 compounds in the 42 CMMs for promoting blood circulation and removing blood stasis were predicted by this model.The key effect features of the Han-Ku-Gan combination were anti-inflammatory,anti-tumor,anti-atherosclerosis,anti-Parkinson,hypoglycemic,and anti-coagulant properties,as well as excitation of uterine smooth muscle.The key effect features of the Wen-Xin-Gan combination were anti-inflammatory,anti-atherosclerosis,anti-hypertensive,anticoagulant,anti-tumor,and anti-cardiac insufficiency effects,as well as enhanced immunity,sedation and hypnosis,and analgesia.Conclusion:This study provides a new method for the further exploration of the relationship between the PCMM and clinical effect features.