The calculation of sensitometric properties of 1,2,4-triazolo[1,5-a]pyrimidines by use of a neural network

Quantitative structure-property relationships were proposed by using artifi cial neural networks and information received from C-13 NMR spectra. The su itability of 1,2,4-triazolo[1,5-a]pyrimidines as stabilizers in photographi c silver halide materials was determined from their chemical structures. Fo r the numeric coding of the chemical structures of differently substituted 1,2,4-triazolo [1,5-a]pyrimidines 1-44 only information available from thei r C-13 NMR spectra was used. Even an assignment of the C-13 NMR chemical sh ift values to the carbons was not necessary. The best results were achieved by combination of the C-13 NMR chemical shifts of carbons of the basic het erocycle and the relative fog D-rel using a feed-forward two-layer neural n etwork. For some compounds with a good stabilizing effect the calculated re sults strongly differ from experimental values giving indication of a mecha nism which is not covered by the C-13 NMR chemical shifts.