Pharmacokinetics of Active Ingredients of Cynanchum Otophyllum in Rat Plasma with Small Samples by Nonlinear Least Squares Estimation and Machine Learning Method

We studied the one compartment pharmacokinetic model of the inhibitory effect of external administration of Cynanchum otophyllum (Co) components M1 and M2 combined with the drug phenobarbital (PHB) on epilepsy, one compartment pharmacokinetic model was obtained by integral solution, and the relationship between plasma concentration and time was obtained by Laplace transform. The inhibitory effect of M1 and M2 on epilepsy was achieved by two methods of extravascular administration, PHB and M1+M2+PHB in these two groups of experiments to obtain the corresponding plasma concentration and maximum electrical shock data. The experimental data were enhanced by synthetic minority oversampling technique (SMOTE). According to the observed data, SPSS nonlinear least squares fitting principle was used to estimate the relationship function parameters between plasma concentration and time, and R2 was used to evaluate its effectiveness. In order to further verify the pharmacokinetic and pharmacodynamic models, we used the traditional machine learning method support vector machine (SVM) to train the samples to test the anti-epileptic effects of Co components M1 and M2. 63 samples were divided into training group and test group according to 7:3. The training used polynomial function kernel and loss function used cross entropy function. By using SPSS Modeler, the results of test group before and after the test group enhancement are 52.6% and 60.2%. The comparison result shows that the administration mode of M1+M2 can assist PHB to treat epilepsy. As a result, it could try a variety of options for the country's safe medication and clinical optimization.