Classificação de analgésicos utilizando técnica espectroscópica e termoanalítica associadas a métodos quimiométricos

Abstract

In the last years occurred a significant increase in the use of analgesics, for example, those that contain acetaminophen as the active pharmaceutical ingredient, being essential for the pharmaceutical industry and the supervisory organs a rigorous control in the production of these medicines. Thus, it becomes necessary the improvement of techniques with the application of reliable analytical methodologies, which are, preferably, quick and low cost, as, for example, the Near Infrared (NIR) Spectroscopy and the Differential Scanning Calorimetry (DSC). Though, even being techniques with extensive analytical power, its use is hampered in samples such as medicines, because results are presented in a complex way for direct interpretation, making the use of chemometric methods necessary. In this context, the objective of this study was to analyze by differential scanning calorimetry and by near infrared spectroscopy, combined with multivariate chemometric techniques, medicines containing acetaminophen and caffeine. So, three brands of medicines were analyzed by DSC and two classes of medicines by NIR. Having the DSC curves obtained in nitrogen atmosphere (50 mL min -1 ), in the temperature range from 92.00 to 190.00 °C, with heating rate of 10 ° C min -1 and preprocessed with the technique Standard Normal Variate (SNV), and the NIR spectra obtained in the interval from 1950 to 2500 nm and preprocessed employing the first derivative, with the filter Savitzky-Golay, second order polynomial and window of 19 points. Posteriorly, it was performed the analgesics classification using the models Linear Discriminant Analysis (LDA) with Successive Projection Algorithm (SPA) and with Genetic Algorithm (GA-LDA) as variable selection techniques, and the K-Nearest Neighbor (KNN), in addition, the DSC curves data were also submitted to Principal Components Analysis (PCA). It was observed for the data obtained by DSC, that the PCA separated the brand M3 of M1 and M2; SPA-LDA and GA-LDA presented a success rate of 94.74 % for the training set and 90,00 % for the test set and the KNN method classified the samples with 100 % of success. On the other hand, for those obtained by NIR, in the SPA-LDA model the success rate was 97.77 % and 84.44 %, in the GA-LDA 96.66 % and 93.33 %, and in the KNN method 100 % and 80 %, for training set and test set, respectively. Thereby, the analysis of the obtained results showed that DSC and NIR techniques aggregate to chemometric methods are efficient alternatives to the use of High Performance Liquid Chromatography (HPLC), with the advantage of achieving results quickly, low cost and without generating pollutant residues, making feasible the use of these techniques to streamline the quality control in pharmaceutical industries, as well as, to assist the surveillance authorities in the rapid detection of medicines adulteration.