In this study, two chemometric techniques, partial least-square regression (PLSR) and artificial neural network (ANN) were developed and compared for the simultaneous assay of paracetamol (PCT) and caffeine (CAF) in pharmaceutical formulations by using spectrophotometric data. Six different concentrations of paracetamol and caffeine were considered to make mixture solutions of standard samples by using orthogonal experimental design (OED). UV spectra of these mixtures were recorded in the wavelength range of 205-300 nm versus a solvent blank and digitized absorbance was sampled at 1 nm intervals. Drug concentrations and instrumental spectra of 36 mixture solutions were used for model development and validation and finally 6 commercially available tablets were used to test the developed models. ANN shows better prediction efficiencies than that of PLSR with R2 value 99.28% for prediction and 99.13% for validation set. These two models were successfully applied to commercial pharmaceutical formulations, and it is found by ANN that the drugs contain 75 to 86% of paracetamol and 77 to 92% of caffeine of their label claim. Either of the proposed methods is simple and rapid and can be easily used in the quality assessment of drugs as an alternative analytical method. Bangladesh J. Sci. Ind. Res.54(3), 215-222, 2019

There is a large variety and trademarks of vegetable oils in Bangladesh. The oils have characteristics very similar to each other and often cannot be classified by the observation of color, odor or taste. This paper proposes a vibrational spectroscopic method like FTIR in association with chemometric techniques to classify vegetable oils like: sunflower, mustard, sesame, soybean, castor, olive and palm oils from different manufacturers. In the FTIR spectra of oil, as information about fatty acid composition is concentrated in the range of 4400-200 cm-1 principal component analysis (PCA) was applied on the standardized full FTIR spectral data of this region for vegetable oils to totally capture the FTIR spectral pattern; seven varieties of vegetable oils could be successfully classified from their PCA scores. PCA of FTIR spectra of different known vegetable oils is used to determine the identity of several unknown vegetable oils. The unknowns are then analyzed, plotted, and identified based on their proximity to the known in principal component space. For the multivariate analysis PCA and soft independent modeling of class analogy (SIMCA) and support vector machines (SVMs) were used. 85% and 14% variability of data was explained by PC1 and PC2 respectively. PC1 has strong positive correlation with soybean, sunflower, palm and olive oil while strong negative correlation with mustard, castor and sesame oil. Soybean oils are positively and sesame oils are negatively correlated with PC2. Unknown oil samples can be identified properly by used supervised methods i.e. SIMCA, SVM by developing model with the help of PCA. The major interest of this method using chemometric analysis of spectral data is in their rapidity, since no chemical treatment of samples is required.

No abstract