Abstract Background Polygonatum cyrtonema Hua (PCH) is widely used in traditional Chinese medicine. The components and activity are important for simultaneously quality evaluation in PCH. Objective This study aimed to establish an accurate, rapid and comprehensive quality evaluation method by near-infrared spectroscopy (NIR) for forecasting the total polysaccharides content (TPsC), total flavonoids content (TFC), and antioxidant activity (AA) in PCH. Methods PCH samples were used to establish model of TPsC, TFC, and AA in PCH by NIR combined with partial least squares regression (PLS). To enhance the accuracy of the models and remove non-essential variables, we used multiple spectral preprocessing methods and multiple chemometrics to analyze the processed full spectrum, such as competitive adaptive reweighted sampling-partial least squares regression (CARS-PLS), moving window-partial least squares regression (MW-PLS), and interval random frog-partial least squares regression (iRF-PLS). The remaining samples were used to complete external validation. Results Satisfactory prediction results of PLS models combined with chemometrics were obtained. The optimal chemometric of TPsC, TFC, and AA selected 157, 105, and 134 variables, respectively. For the TPsC, TFC, and AA models, which incorporated optimal spectral preprocessing and chemometric, the root mean square error of calibration (RMSEC, %) values were 0.743, 0.069, and 0.136, while the R2 values were above 0.95. The root mean square error of prediction (RMSEP, %) for these models were 1.00, 0.074, and 0.153, while R2 values were above 0.90. The remaining 20 samples were used to complete external validation, which confirmed the preeminent comparability and ability of the proposed method. Conclusion NIR combined with chemometrics provide an effective, fast, and nondestructive approach for evaluating quality via multiple indicators of PCH and provides a meaningful reference for evaluation of herbal medicine quality. Highlights NIR combined with chemometrics offers an effective, quick, nondestructive way to evaluate herbal medicine quality with multiple indicators.
Fan et al. (Thu,) studied this question.