Abstract
A comparative study is established among 4 chemometric models depending on classical least squares (CLS) approach, namely, spectral residual augmented CLS (SRACLS), net analyte processing CLS (NAP-CLS), orthogonal signal correction CLS (OSC-CLS), and direct orthogonal signal correction CLS (DOSC-CLS). The comparison is expressed through analysis of a case study dataset of UV spectral data of Cefoperazone Sodium (CEF) and its two related impurities: in pure powder form and in pharmaceutical dosage form. Four-level three-factor experimental design was established for optimum analysis. The adopted experimental design gave rise to a training set consisting of 16 mixtures (containing different ratios of interfering species). To test the prediction power of the suggested models, an independent test set consisting of 9 mixtures was used. The presented results show the ability of the proposed models to quantify CEF in presence of two related impurities with high accuracy and selectivity (103.76 +/- 1.03, 102.07 +/- 0.91, 101.61 +/- 0.72, and 101.60 +/- 0.72 for SRACLS, NAP-CLS, OSC-CLS, and DOSC-CLS, resp.). Dosage form analysis results were compared statistically to a published HPLC methodology showing insignificant difference in terms of precision and accuracy, indicating the suggested models reliability and their suitability for quality control analysis of drug product. Compared to other models, OSC-CLS and DOSC-CLS models gave more accurate results with lower prediction error for test set samples.