The purpose of this study was to construct a calibration model for the prediction of glycyrrhizic acid content in Kakkonto extracts using near infrared (NIR) spectroscopy. The NIR spectra of the Kakkonto extracts were obtained using a Fourier transform NIR spectrometer in transmission mode and chemometric analysis was performed using partial least-square (PLS) regression. The calibration model was constructed by the selection of wavenumber regions and by the first derivative pre-treatment of NIR spectra. The glycyrrhizic acid content could be predicted using a calibration model comprising three principal components (PCs) obtained by the PLS method. The calibration model was theoretically analysed by investigating the standard error of prediction values, the loading vectors of each PC and the regression vector. The relationship between the actual and predicted glycyrrhizic acid contents in the Kakkonto extract exhibited a straight line with a coefficient of determination of 0.966 (calibration) and 0.945 (validation), respectively. The predicted glycyrrhizic acid content in the Kakkonto extract was within the 95% predictive intervals.

Keywords: partial-least squares regression (PLS), Kakkonto extract, glycyrrhizic acid, process analytical technology (PAT)