Quantitative and qualitative prediction of corneal permeability for drug-like compounds

► Linear and nonlinear prediction of corneal permeability using molecular descriptors. ► MLP-NN model was found to be more successful than MLR equation. ► Molecular volume and molecular weight were identified as the most important descriptors. ► Categorizing drugs in two low permeable and very low permeable compounds groups. ► CPNN model can correctly recognize objects belonging to the groups.