Development of a rapid quality assessment technique for Radix Paeoniae Alba (Paeonia lactiflora Pall.) using near-infrared spectroscopy and chemometrics analysis

Abstract

Radix Paeoniae Alba (RPA) was subjected to a quick quality evaluation procedure using near-infrared (NIR) spectroscopy and chemometrics. The orthogonal partial least squares discrimination analysis (OPLS-DA) method was applied to the spectrum analysis based on SIMCA software, and a qualitative discriminant model was constructed to differentiate between the origin of RPA. Additionally, the NIR spectroscopy quantitative analysis models of gallic acid, methyl gallate, oxypaeoniflorin, catechin, albiflorin, paeoniflorin, 1,2,3,4,6-O-pentagalloylglucose, and benzoylpaeoniflorin were established by partial least squares method, with the content of components determined by HPLC serving as the reference value. To select the optimal spectroscopy pretreatment technique, the correlation coefficient R, root mean square error of calibration, root mean square error of prediction, and performance index were employed as assessment indices. The variable importance projection map was created using the OPLS-DA method to maximize the detection spectral band. The optimal number of factors was then determined using cross-validation, using the anticipated residual error sum of squares and the root mean square error of cross-validation as indicators. Ultimately, a quantitative model of the NIR spectrum was established using partial least squares with the spectral area of 9997.17 ∼ 8612.53 cm⁻¹. Standard normal variation, second derivative, and no smoothing were used as pretreatments for the spectrum. The correlation coefficients of the eight components were all over 0.99, according to the model. Rapid, stable, dependable, and free of chemical reagent usage are the characteristics of the qualitative and quantitative models created in this work, which can be applied to the quick assessment of RPA's quality.