A Linear Principal Component Regression and Nonlinear Neural Network Model for Determination of Indomethacin in Plasma Samples Using UV-Vis Spectroscopy and Comparison with HPLC

A sensitive and selective method using combination of two chemometrics methods, principal component analysis (pca) and artificial neural network (ann), and uv-visible spectroscopy has been developed for the determination of indomethacin (idm) in plasma samples. initially the absorbance spectra were processed using pca to noise reduction and data compression. the scores of these pcs were used as the inputs of ann. the ann trained by the back-propagation learning was employed to model the complex non-linear relationship between the pcs extracted from uv-visible spectra of idm and the absorbance values. nonlinear method (pc-ann) was better than the pcr method considerably in the goodness of fit and predictivity parameters and other criteria for evaluation of the proposed model. optimal ann model were as follows: number of input pcs: 2, number of neurons in hidden layer: 3. the linear calibration range was 1×10^-7 to 2.4×10^-6 m, the detection limit were 0.21 × 10^-7 m., the results have been compared with those obtained by the hplc method.