Predicting the Anticonvulsant Activities of Phenylacetanilides Using Quantitative-structure-activity-relationship and Artificial Neural Network Methods

In this study, anticonvulsant activity of phenylacetanilides compounds was predicted using QSAR and artificial neural network (ANN) models. Variety kinds of molecular descriptors were computed using Dragon for 30 monosubstituted phenylacetanilides. Then, seven out of 1600 descriptors were selected and used in ANN analysis. The complete set of 30 compounds was randomly divided into a training set of 80%, a test set of 10%, and a validation set of 10% compounds. Moreover, multiple linear regression (MLR) analysis was utilized to build a linear model by using the same descriptors and the results of this linear model were compared with the nonlinear ANN analysis. The obtained Correlation coefficient (R2) and mean squared error (MSE) of the ANN and MLR models (for the whole dataset) were 0.85, 0.06816; and 0.6, 0.09792, respectively. The higher R2 of ANN method revealed that the relationship between the descriptors and anticonvulsant activity of the compounds is non-linear.