Simultaneous kinetic spectrophotometric determination of acetaminophen andphenobarbital by artificial neural networks and partial least squares

A kinetic spectrophotometric method has been developed for simultaneous qua ntitative determination of acetaminophen and phenobarbital in pharmaceutica l preparations, and relies on the different kinetic rates of the analytes i n their oxidative coupling reaction with 3-methylbenzothiazolin-2-one hydra zone (MBTH) in the presence of hydrochloric acid and the Fe(III) oxidant. T he absorbance was measured and recorded from 440 to 740 nm every 60 s from 15 to 435 s. An artificial neural network (ANN) coupled with principal comp onent analysis (principal component artificial neural network, PC-ANN) has been described, and its principles and applications were illustrated by a s imulated kinetic system. The performance of the PC-ANN was compared with a conventional ANN model. PC-ANN, as well as ANN and partial least squares (P LS), were applied to a validation set of overlapping spectra containing kin etic data of binary mixtures with different concentration ratios of acetami nophen and phenobarbital. The results show that PC-ANN is the most efficien t of the three chemometrics methods applied for prediction of the two analy tes. (C) 2000 Elsevier Science B.V. All rights reserved.