QSRR Prediction of Immobilized Artificial Membrane Retention Factors of Some Drugs

In this work multiple linear regression (MLR) and artificial neural network (ANN) were used to predict the retention factors of 40 basic and neutral drugs in immobilized artificial membrane liquid chromatography. Two separate models were developed for prediction of solute retention in two mobile phase compositions which were used five identical descriptors. The standard errors in ANN calculation of for training, internal and external test sets were 0.205, 0.3299 and 0.389, respectively, while these values are 0.280, 0.426 and 0.448, respectively for MLR model. Also the standard errors in ANN prediction of for training, internal and external test sets were 0.144, 0.596 and 0.557, respectively, while these values are 0.318, 0.613 and 0.453, respectively for MLR model. The validation and robustness of these ANN models were evaluated by cross-validation and Y-scrambling methods, which produce successful results.