RESUMO
The present paper proposes an analytical method for fast near-infrared (NIR) determination of dipyrone in injectable formulations with a nominal content of 50.0%mv(-1) without violation of the ampoule. For this purpose, two multivariate calibration methods are evaluated, namely Partial-Least-Squares (PLS) and Multiple Linear Regression (MLR) with variable selection by the Successive Projections Algorithm (SPA). The resulting models comprised four latent variables (PLS) and five spectral variables (MLR-SPA). Appropriate predictions were obtained in both cases, with RMSEP values of 0.39 (PLS) and 0.35%mv(-1) (MLR-SPA) and correlation coefficients of 0.9970 (PLS) and 0.9975 (MLR-SPA) for a calibration range of 40-60%mv(-1). No systematic error was observed and no significant differences were found between the predicted and reference values, according to a paired t-test at 95% confidence level.
