Quantitative Analysis of Four Active Components in Wine-Processed Scutellaria baicalensis by Data Fusion Strategy Coupled with Multivariate Statistical Analysis

Wine-processed Scutellaria baicalensis is a processed Chinese herbal product obtained by wine-processing of raw S. baicalensis. Flavonoids are its core pharmacodynamic constituents with heat-clearing, anti-inflammatory, hepatoprotective and other bioactivities. High-performance liquid chromatography (HPLC) remains the reference for their quantification but is slow and solvent-intensive. This study proposes a chromatography-free alternative that fuses near-infrared (NIR) and ultraviolet (UV) spectra through low- and mid-level data fusion, coupled with interval partial least squares (iPLS) variable screening, to quantify the four flavonoids simultaneously across five processing degrees (raw plus four stir-frying durations, n = 90). HPLC values served as reference and the set was split by the Kennard–Stone algorithm. Against single-spectroscopy models, data fusion improved prediction for all four analytes: low-level fusion benefited the aglycones baicalein and wogonin (e.g., wogonin: Rc2 = 0.973, Rp2 = 0.979), whereas mid-level fusion raised the glycosides baicalin and wogonoside to satisfactory accuracy. Across all analytes, calibration R2 > 0.93, prediction R2 > 0.90, and RPD > 2.5. The method supports rapid, non-destructive evaluation of wine-processed S. baicalensis and is extensible to other processed herbal medicines.