Neural network method to estimate the aqueous rate constants for the OH reactions with organic compounds

The development of a structure/reactivity relationship for aqueous phase reactions of organic compounds with OH is presented. An empirical model is used to relate the functional groups of the molecule and the rate constant of the molecule reacting with OH. An artificial neural network, the multi layer perceptron (MLP), is found to be capable of predicting the rate constants, for which the slope of the modeled vs. experimental data, the R2 and the F-test are statistically significant at the 95% confidence level. The MLP estimates 87% of the predicted data within a factor of 2 of the experimental data. The standard error of prediction of the logarithm of the reaction rate constants is 0.24 for a logarithm of the reaction rates ranging from 7 to 10.