Abstract
Five different chemometric methods were developed for the simultaneous determination of betamethasone dipropionate (BMD), clotrimazole (CT) and benzyl alcohol (BA) in their combined dosage form (Lotriderm® cream). The applied methods included three full spectrum based chemometric techniques; namely principal component regression (PCR), Partial Least Squares (PLS) and Artificial Neural Networks (ANN), while the other two methods were PLS and ANN preceded by genetic algorithm procedure (GA-PLS and GA-ANN) as a wavelength selection procedure. A multilevel multifactor experimental design was adopted for proper construction of the models. A validation set composed of 12 mixtures containing different ratios of the three analytes was used to evaluate the predictive power of the suggested models. All the proposed methods except ANN, were successfully applied for the analysis of their pharmaceutical formulation (Lotriderm® cream). Results demonstrated the efficiency of the four methods as quantitative tool for analysis of the three analytes without prior separation procedures and without any interference from the co-formulated excipient. Additionally, the work highlighted the effect of GA on increasing the predictive power of PLS and ANN models.
The absorption spectra of BMD 8μg/mL (ـــــ), CT 80μg/mL (ـــــ) and BA 120μg/mL (ـــــ) in methanol [Display omitted]
•No chemometric method has been reported for the mixture determination.•Applying PLS, PCR and ANN to resolve severely overlapped spectral data•Application of GA prior to PLS and ANN chemometric models•Evaluation of the effect of GA as a powerful variable selection tool•Statistical comparison to investigate the effect of GA on each model