Desenvolvimento de uma metodologia para o controle de qualidade de medicamento injetável, sem violação da ampola, usando espectroscopia NIR e técnicas quimiométricas

Abstract

In this work, a new methodology is proposed that uses a combination of Near Infrared Spectroscopy (NIR) with chemometric methods of multivariate analysis for the determination of active ingredient in parenteral drug. The drug used in this study was dipyrone. To create the models of multivariate calibration, standard samples of dipyrone prepared in laboratory and commercial samples were used. NIR spectra were recorded, pre-processed and used to build models of pattern recognition and multivariate calibration. For the initial analysis of data, an exploratory analysis was performed from PCA models. Then, SIMCA models were developed and used for Screening analysis of the samples, at a confidence level of 95%. These tests served to confirm which set of samples (commercial and prepared in the laboratory) could be used in the composition of sets of calibration, validation and prediction employed in the development of the methodology. For the selection of samples from the sets we used the SPXY algorithm. Initially, the calibration set containing commercial and prepared in the laboratory samples was used to build PLS1, MLR-SPA, SW-MLR, MLR-ASA and MLR-GA multivariate calibration models. These models were validated using a test set and applied to the determination of dipyrone using a prediction set. The results showed that all models presented satisfactory prediction errors with respect to the reference method, and it is very promising for predicting the dipyrone content in drugs injection. As the proposed methodology uses a non invasive process it is not necessary a previous treatment of the samples, no reagent was used, and a relatively high analytical rate is achieved.