Multivariate calibration: a powerful tool in pharmaceutical análisis

P. DAMIÁNI, G. ESCANDAR, A. OLIVIERI* AND H.C. GOICOECHEA*

Current Pharmaceutical Analysis

Benthan

Lugar: Karachi; Año: 2005 vol. 1 p. 145 - 154

1573-4129

The field of pharmaceutical analysis calls for rapid, simple and accurate methods that ensure the quality of the products. The determination of the content of active compounds in pharmaceutical preparations is often made by using high-performance liquid chromatography (HPLC). However, this technique has some disadvantages, such as the time required to perform the analysis. On the other hand, multivariate calibration methods are being successfully applied to solve a wide variety of tasks in different fields of analytical chemistry. Several of these methods have been presented in the literature in the last years. Among them, partial least squares (PLS) has become a routine practice in pharmaceutical analysis. However, there are certain problems that require more sophisticated models to be solved. We present several situations in which conventional multivariate models have been successfully applied to simultaneously analyze several active principles in tablets, syrups, injections and drops. On the other hand, we also present events in which the presence of non-linearity was taken into account by using artificial neural networks (ANNs). Finally, a new tool that allows one to analyze samples containing non modelled interferences is shown, based on the acquisition of matrix-type (second-order) data, and subsequent construction of models presenting the well known “second order advantage”