NEURAL NETWORK STUDIES .1. ESTIMATION OF THE AQUEOUS SOLUBILITY OF ORGANIC-COMPOUNDS

A study has been made of the ability of neural networks to estimate the aqueous solubility of a wide range of organic compounds. A training set of 331 compounds was used and the trained neural network tested on a prediction set of 19 unknown compounds. A comparison was made with the results obtained from a previous study using regression analysis (Bodor et al. 2). On the basis of training set results, the neural network model exceeded the performance of the regression analysis technique and gave a prediction which was sufficiently accurate to estimate the water solubility of an organic compound based entirely on calculated values of selected properties.