Prediction of selectivity coefficients of a theophylline-selective electrode using MLR and ANN

The selectivity coefficient of 24 interfering compounds (drugs, amino acids and organic compounds) of a theophylline-selective electrode was predicted using an artificial neural network (ANN). The multiple linear regression (MLR) technique was used to select the descriptors as inputs for the artificial neural network. The neural network employed here is a connected back-propagation model with a 2-2-1 architecture. Two topological indices for the interfering compounds, namely, Narumi harmonic topological index, HNar, and sum of topological distances between nitrogen and oxygen, T(N⋯O), were taken as inputs for the ANN. Standard errors of training and prediction were 0.954 and 0.945, respectively, for the MLR model and 0.032 and 0.007, respectively, for the ANN model. Two topological indices for the interference of the electrode were taken as inputs for ANN.