Near-infrared spectroscopy is evaluated as an on-line technique for monitoring the homogeneity of a pharmaceutical blend during the blending process. Blends containing 10% sodium benzoate (model active), which provided an aromatic functionality typical of many pharmaceutical compounds, 39% microcrystalline cellulose (Avicel PH102), 50% lactose, and 1% magnesium stearate were developed to mimic the properties of an actual pharmaceutical blend. A twin-shell V-blender was modified to allow installation of a diffuse reflectance fiber-optic probe at the axis of rotation, and spectra were collected during three experiments using a commercially available near-infrared spectrophotometer. In each experiment, blender control and spectral data collection were controlled by a compilation of software packages. The experiments detected spectral changes which eventually converged to a point of constant variance. Further analysis of the spectral data showed the blend is homogeneous long before a typical blending period is complete. Near-infrared spectroscopy has proven to be a feasible and effective method for the "real time" noninvasive determination of homogeneity in a pharmaceutical blend.

No abstract

SUMMARYThis tutorial paper presents a simplified view of one of the more recently published multivariate calibration methods particularly suited to dealing with non-linear data sets. The method is referred to as MARS and stands for multivariate adaptive regression splines. Simple examples are provided to explain the workings of the method.

No abstract

No abstract