Licorice is a medicine-food dual-purpose plant, yet reliable quality assessment of its processed forms remains challenging because external phenotypic and spectral signals are rarely linked to validated chemical markers within a unified framework. An integrated "trait–spectrum–chemical marker" framework was established by combining UV–Vis–NIR multispectral imaging, UPLC–Q-TOF/MS chemical profiling, and machine learning. Six characteristic marker constituents (liquiritin, isoliquiritigenin, glycyrrhizin, 18β-glycyrrhetinic acid, licochalcone A, and formononetin) were confirmed against reference standards, providing a reliable chemical basis. Machine-learning classifiers achieved 90–95% accuracy, with CIELab parameters and key visible–NIR bands (430–450 nm, 590–630 nm) as the most informative predictors, consistent with the absorption properties of the validated constituents. OPLS-DA pairwise comparisons identified isoliquiritigenin, formononetin, liquiritin, and 18β-glycyrrhetinic acid as key discriminant markers, reflecting processing-induced remodeling of the flavonoid–saponin balance. Network analysis revealed quantifiable cross-level associations among phenotypic traits, spectral bands, and validated chemical markers; molecular docking and 100 ns molecular dynamics simulation confirmed the isoliquiritigenin–SRC pair as a representative component-target association. This framework advances quality assessment of processed licorice from empirical classification towards chemically interpretable evaluation, offering a validated non-destructive strategy for food composition analysis and quality standardization. • A cross-level framework integrates phenotype, spectra, and metabolomics. • Multispectral imaging with ML achieves >90% accuracy in process monitoring. • Metabolomics elucidates processing-induced chemical transformations. • Network analysis bridges the gap between phenotype and functional targets.
Huang et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: