Development of validated QSPR models for O-H bond dissociation energy in substituted phenols

作者：Xu, Qi; Xu, Jie^*

来源：Monatshefte fur Chemie, 2017, 148(4): 645-654.

DOI：10.1007/s00706-016-1794-7

• Molecular modeling
• QSAR
• Neural network
• Bond dissociation energy
• Substituted phenols
• OECD principles

A quantitative structure-property relationship (QSPR) study was reported to predict the O-H bond dissociation energy in substituted phenols based on a training set of 60 compounds and a test set of 18 compounds divided by DUPLEX algorithm. The QSPR models were built by stepwise multiple linear regression (MLR) and nonlinear artificial neural network (ANN), respectively. The mean relative error (MRE) provided by the MLR model was 1.51 %, indicating satisfactory predictive ability; while the MRE was further improved to 1.28 % by the ANN model. The results from Y-randomization tests, LMO cross-validations, and external validation through test set confirmed the reliability and robustness of the models. Moreover, the applicability domain of the developed models was assessed and visualized by the Williams plot.

• 出版日期2017-4
• 单位武汉纺织大学