QSPR study of supercooled liquid vapour pressures of polybrominated diphenyl ethers using the molecular distance–edge vector index

The quantitative structure property relationship (QSPR) for supercooled liquid vapour pressures (pL) of polybrominated diphenyl ethers (PBDEs) was investigated. The molecular distance–edge vector (MDEV) index was used as the structural descriptor. The quantitative relationship between the MDEV index and log pL was modelled by multivariate linear regression (MLR) and an artificial neural network (ANN). The leave-one-out cross validation and k-fold cross validation were performed to assess the prediction ability of the developed models. For the MLR method, the prediction root mean square relative error (RMSRE) of the leave-one-out cross validation and the k-fold cross validation were 9.95 and 9.05, respectively. For the ANN method, the prediction RMSRE of the leave-one-out cross validation and the k-fold cross validation were 8.75 and 8.31, respectively. It was demonstrated that the established models were practicable for predicting the log pL values of PBDEs. The MDEV index was quantitatively related to the log pL of PBDEs. MLR and linear ANN were practicable for modelling this relationship. Compared with the MLR, the ANN method exhibited slightly higher prediction accuracy. Subsequently, an MLR model, the regression equation of which was log pL = = 0.2868M11 – 0.8449M12 – 0.0605, and an ANN model, which was a two-input linear network, were developed. The two models could be used to predict the log pL value of each PBDE.