On Pharmacokinetics and Pharmacodynamics of Antiepileptic Components of Qingyangshengenin Based on SVM and LSTM

A compartmental pharmacokinetic model of Qingyangshengenin (QYS) components M1 ((Qingyangshengenin 3-0- β -D-oleandropyranosy1 (1→4)- β -D-Ganadamaropyranosy1 0-(1→4)- β -D-Digitoxin pyranoside)) and M2 ((Qingyangshengenin 3-0-β -D-oleandropyranosy1-(l→4)- β -D-Ganadamaropyranosy1 (1 → 4)- β -D-Digitalis pyranosyl-(l → 4)- β -D-Ganadamaropyranoside)) combined with normal antiepileptic drug Phenobarbital (PHB) was investigated on the effectiveness inhibitory of epilepsy. The model was derived from the differential solution firstly, and then the relationship between plasma concentration and time was obtained. Rats were administrated with PHB and PHB+M1+M2 respectively. And maximal electroconvulsive shock (MES) was used to judge the anti-epileptic effect of M1 and M2. Based on the observed data, parameters of relationship function between plasma concentration and time were estimated by using nonlinear least squares fitting principle. Taking PHB group as a benchmark, area under the drug time curve (AUC) of two administrations (PHB, PHB+M1+M2) were calculated to determine whether M1 and M2 were the active components for suppressing epilepsy. Convolution of effect site distribution function and pharmacokinetic model combined with logistic regression was used to calculate the anti-epilepsy scores of metabolites group (CPHB and CPHB+M1+M2). For further validation of pharmacokinetic model, Support Vector Machine (SVM) and Long Short-Term Memory (LSTM) were used to train samples for testing the antiepileptic effectiveness of QYS. The results show M1 and M2 are antiepileptic components.