In this study, utilization of artificial neural network (ANN) models [static-multilayer perceptron (MLP) and generalized regression neural networks and dynamic-gamma one-layer network and recurrent one-layer network] for prediction of diclofenac sodium (DS) release from drug-cationic surfactant-modified zeolites physical mixtures comprising different surfactant/drug molar ratio (0.2-2.5) was performed. The inputs for ANNs trainings were surfactant/drug molar ratios, that is, drug loadings in the drug-modified zeolite mixtures, whereas the outputs were percents of drug release in predetermined time points during drug release test (8 h). The obtained results revealed that MLP showed the highest correlation between experimental and predicted drug release. The safety of both natural and cationic surfactant-modified zeolite as a potential excipient was confirmed in an acute toxicity testing during 72 h. DS (1.5, 5, 10, mg/kg, p.o.) as well as DS-modified zeolites mixtures produced a significant dose-dependent reduction of the rat paw edema induced by proinflammatory agent carrageenan. DS antiedematous effect was intensified and prolonged significantly by modified zeolite. These results could suggest the potential improvement in the treatment of inflammation by DS-modified zeolite mixtures.
Keywords: adsorption; antiedematous activity; cationic surfactant; clinoptilolite; diclofenac sodium; dissolution; dose-response; excipient; in silico modeling; neural networks.
© 2014 Wiley Periodicals, Inc. and the American Pharmacists Association.