A robust quantitative near infrared modeling approach for blend monitoring


    loading  Checking for direct PDF access through Ovid

Abstract

HighlightsA Near-infrared calibration strategy was developed for pharmaceutical blend monitoring in which only gram scale material was required to monitor kilogram scale blending process.Quantitative Near-Infrared spectroscopy models for blend monitoring developed on small-scale with same optical setup and no blender showed similar performance to models built on larger scale with the blender.Density control on small-scale showed to have significant effect when modeling on larger scale as the prediction bias increased when density was not controlled in the small-scale calibration dataset.This study demonstrates a material sparing Near-Infrared modeling approach for powder blend monitoring. In this new approach, gram scale powder mixtures are subjected to compression loads to simulate the effect of scale using an Instron universal testing system. Models prepared by the new method development approach (small-scale method) and by a traditional method development (blender-scale method) were compared by simultaneously monitoring a 1 kg batch size blend run. Both models demonstrated similar model performance. The small-scale method strategy significantly reduces the total resources expended to develop Near-Infrared calibration models for on-line blend monitoring. Further, this development approach does not require the actual equipment (i.e., blender) to which the method will be applied, only a similar optical interface. Thus, a robust on-line blend monitoring method can be fully developed before any large-scale blending experiment is viable, allowing the blend method to be used during scale-up and blend development trials.

    loading  Loading Related Articles