Desenvolvimento de modelos de regressão multivariada para a quantificação de benzoilmetronidazol na presença de seus produtos de degradação por espectroscopia no infravermelho próximo

Abstract

Benzoyl metronidazole (BMZ) is a drug with antiparasitic and antibacterial activity available in the form of pediatric suspensions. The BMZ main degradation products are metronidazole and benzoic acid, and there are no reports in the literature on the determination of BMZ in the presence of its degradation products using near infrared spectroscopy. Therefore, in this study a method for determining the content of BMZ pharmaceutical ingredient in the presence of its main degradation products by near infrared spectroscopy associated with multivariate calibration were to develop. Regression with variable selection methods such as partial least squares regression for interval (iPLS) and partial least squares regression for synergism intervals (siPLS) were applied in order to select spectral regions that produce models with smaller errors. The best model using the iPLS algorithm was obtained when the spectrum was divided into 12 sub-intervals and select a period 11 (RSEP% = 1.37). Once the spectrum has been divided into 16 intervals and combined subintervals 9, 13 and 18 yielded the best model for siPLS algorithm (RSEP = 1.30%). The proposed method can be considered selective; it allows determining the BMZ in the presence of its degradation products.