Structure-based predictions of 1H NMR chemical shifts using feed-forward neural networks

Feed-forward neural networks were trained for the general prediction of H-1 NMR chemical shifts of CH, protons in organic compounds in CDCl3. The training set consisted of 744 H-1 NMR chemical shifts from 120 molecular structures. The method was optimized in terms of selected proton descriptors (selection of variables), the number of hidden neurons, and integration of different networks in ensembles. Predictions were obtained for an independent test set of 952 cases with a mean average error of 0.29 ppm (0.20 ppm for 90% of the cases). The results were significantly better than those obtained with counterpropagation neural networks.