Simultaneous multi-component quantification and geographical origin discrimination of Taraxaci Herba using near-infrared spectroscopy combined with chemometrics.

This study aimed to employ near-infrared spectroscopy (NIRS) combined with chemometric methods for the rapid quality assessment of Taraxaci Herba (TH). First, different batches of TH samples were collected, and the content of chicoric acid (CHA), caffeic acid (CA), and luteolin (LUT) in TH were determined by high-performance liquid chromatography (HPLC), while total flavonoids (TFs) and total phenols (TPs) were measured using ultraviolet-visible (UV-Vis) spectrophotometry. Second, NIR spectra were acquired using a near-infrared spectrometer, followed by outlier removal and spectral pretreatment. Specifically, informative wavelengths were selected using genetic algorithm (GA), grey wolf optimization (GWO), and a modified global flower pollination algorithm (mgFPA). Finally, partial least squares regression (PLSR) was applied to quantitatively model the relationship between the content of the aforementioned components and the NIR spectra. For qualitative analysis, the unsupervised method of principal component analysis (PCA) and the supervised methods - partial least squares-discriminant analysis (PLS-DA) and k-nearest neighbors (KNN) - were employed for geographical origin discrimination. As a result, the optimized PLSR models demonstrated excellent predictive performance for all five analytes (R2 > 0.90, RPD > 3). Additionally, the geographic origins of TH samples were successfully classified with 100% accuracy. Thus, the present approach provides an efficient and robust method for the quality assessment and control of TH samples.