QSPR modeling of decomposition temperature of energetic cocrystals using artificial neural network

Abstract

The quantitative structure–property relationship for the decomposition temperature (Td) of energetic cocrystals was investigated. The artificial neural network (ANN) model was employed to predict the Td of cocrystals by using molecular descriptors achieved from Dragon software as input variables. The complete set of 30 cocrystals was randomly divided into a training set of 19, a test set of 6, and a validation set of 5 compounds. Average absolute relative deviations and correlation coefficient (R2) of the ANN model (for the whole dataset) were 1.94% and 0.9784, respectively, indicating satisfactory predictive ability and reliability of the model. Moreover, these data were analyzed by multiple linear regression (MLR) method which showed R2 = 0.7438. The result of MLR proves the necessity of applying ANN models for the prediction of the Td of cocrystals.