A neural network approach based on gold-nanoparticle enzyme biosensor

Gold nanoparticles have demonstrated to be a very useful material for the construction of stable and sensitive glucose oxidase (GOx) amperometric biosensors. However, as for other enzyme electrodes, the lack of specificity for glucose limits their practical applications. Coupling biosensor responses with chemometric tools can be used to solve complex analytical signals from mixtures of species with similar properties. In this work, an amperometric biosensor based on a colloidal gold - cysteamine - gold disk electrode with the enzyme GOx and a redox mediator, tetrathiafulvalene (TTF), co-immobilised atop the modified electrode, was used for the simultaneous determination of glucose and its common interferences, ascorbic acid and uric acid, in mixtures. Analytical data obtained from cyclic voltammograms generated with the biosensor were processed using an artificial neural network (ANN), and the separate quantification of the analytes over a range of 0.1-1 mM each was performed without any pretreatment. In all cases, the correlation coefficients obtained were higher than 0.99 and the mean prediction error was less than 1.7%. Copyright (c) 2007 John Wiley & Sons, Ltd.