Interval‐Partial Least Squares to Eliminate Spectral Moisture Interference in Near Infrared Transmission Quantification of the Active Pharmaceutical Ingredient in Escitalopram Tablets

ABSTRACT Near infrared spectroscopy is a technique commonly applied to pharmaceutical tablets to quantify the active pharmaceutical ingredient (API). During its lifetime, the tablet product may absorb water, which can result in deteriorating performance of the initial calibration model. However, through variable subset reduction, it is possible to avoid the spectral interference from water uptake in the tablets. Calibration models based on samples with lower water uptake were used to predict samples with higher water uptake. Reducing the wavelength range to only the range for which the API absorbs resulted in a prediction error of 0.27% (w/w) compared with the prediction error of 0.29% (w/w) for the full spectral range.