CNS permeability of drugs predicted by a decision tree

To predict the ability of drug-like molecules to penetrate the Central Nervous System (CNS), a decision tree was generated. This algorithm was designed to make a straight forward yes/no decision about the permeability of the blood-brain barrier for a given substance, based on the numerical criteria of a large variety of molecular descriptors. The decision tree achieved a prediction accuracy of 96% for the 186 compounds of the training set and 84% for the test set comprising 38 molecules. We found that CNS+drugs are predicted with a higher accuracy (> 94%) than CNS-substances (> 89%).