Lubricant-Sensitivity Assessment of SPRESS® B820 by Near-Infrared Spectroscopy: A Comparison of Multivariate Methods

Abstract

The predictability of multivariate calibration models, calculated with offline near-infrared spectroscopy (NIRS), assessing impact of magnesium stearate (MgSt) fraction, blending time, and compression force on the tablet breaking force (TBF) of SPRESS B820 was statistically compared. Tablets of lubricated SPRESS B820 were prepared by varying lubrication and compression conditions using 2 full factorial design. Tablets were scanned in reflection mode on a benchtop NIRS. A qualitative principal component analysis and quantitative principal component regression (PCR) and partial least square (PLS) regression relationship between lubricant concentration, blending time, compression force, preprocessed NIR spectra, and measured TBF was calculated with calibration data set. The predictability of calibration models was validated with independent data set. Expected qualitative correlations between MgSt blending time and TBF and a nonlinear relationship between MgSt fraction and TBF were observed. Predictability of PLS comprehensive (0.25%-1% w/w MgSt) model was significantly different from individual 0.25%, 0.5%, and 1.0% w/w MgSt PLS models. In addition, PLS calibration models' predictability was different from PCR calibration models. Thus, added lubrication fraction and adopted multivariate methodology should be selected carefully during the calibration and validation stages as it may have a significant impact on the predictability of the developed models.

Emerging technologies for the non-invasive characterization of physical-mechanical properties of tablets

“…The NIR spectra, however, are broad and consist of overlapping and fused peaks. Thus, in order to extract meaningful information, NIR spectra require systematic data processing with the use of Chemometrics (Dave et al, 2015;Dave et al, 2017). Moreover, development of robust multivariate calibration models using large datasets, and validation of the developed models is required to reliably use NIRS for the qualitative and quantitative examination of materials.…”