Improved prediction of tablet properties with near-infrared spectroscopy by a fusion of scatter correction techniques

Near-infrared (NIR) spectra of pharmaceutical tablets get affected by light scattering phenomena, which mask the underlying peaks related to chemical components. Often best performing scatter correction technique is selected from a pool pre-selected techniques. However, data corrected with different techniques may carry complementary information, hence, use single sub-optimal. In this study, aim prove that NIR models pharmaceuticals can directly benefit fusion information extracted multiple To perform fusion, sequential and parallel pre-processing approaches were used. Two open source sets used for demonstration where assay uniformity active ingredient (AI) content prediction was aim. As baseline, approach compared partial least-squares regression (PLSR) performed on standard normal variate (SNV) data, commonly technique. The results suggest extract their essential obtain high-performance predictive models. error bias reduced up 15 % 57 respectively, PLSR SNV data.