A novel approach for rapid quantification of food supplement components using NIR spectroscopy and spectral data transfer

Abstract

The global interest in food supplements is increasing, creating a growing demand for efficient and reliable analytical methods to assess mixture compositions. Near-infrared (NIR) spectroscopy has been widely employed for this purpose; however, traditional chemometric approaches, such as partial least squares (PLS), require a large number of calibration samples (typically 100–200), making the process time-consuming. To overcome this limitation, we propose the spectral data transfer (SDT) approach, which corrects calculated spectra derived from pure components to more accurately align with real measured spectra. The method was tested on a four-component food supplement containing Melissa officinalis, Hypericum perforatum, Passiflora incarnata, and L-tryptophan. By implementing SDT, we significantly enhanced the prediction accuracy of PLS models, reducing RMSEP for all components. Before SDT, RMSEP values were 5.26, 7.23, 20.43 and 9.56 for Melissa officinalis, Hypericum perforatum, Passiflora incarnata, and L-tryptophan, respectively, while they were 3.43, 2.03, 2.46 and 0.86 after SDT and preprocessing (2nd derivative) respectively. Validation using HPLC reference analysis confirmed the accuracy, robustness, and repeatability of the proposed method, demonstrating its effectiveness in advancing NIR spectroscopy for mixture analysis.