QSPR Studies on the Aqueous Solubility of Selected PCDD/FS by Using Artificial Neural Network Combined with Principal Component Analysis

A practicable quantitative structure property relationship (QSPR) model for predicting aqueous solubility, S_w, of 23 poly chlorinated dibenzo-p-dioxins and poly chlorinated dibenzofurans (PCDD/Fs) was developed. Linear artificial neural network (L-ANN) was used to develop the calibration model of S_w. The input variables of L-ANN were obtained from 11 structural descriptors of the investigated PCDD/Fs by using principal component analysis (PCA). Leave one out cross validation was carried out to assess the predictive ability of the model. The result of leave one out cross validation is satisfactory. The R² between the predicted and experimental S_w is 0.9631 and the RMS%RE is 4.87 for all the investigated compounds. It is demonstrated that L-ANN combined with PCA is a practicable method for developing QSPR model for S_w of PCDD/Fs. In addition, PCA is shown to be an applicable approach for the generation of input variables when developing an L-ANN model.