Integrated Chiral Volatile Organic Compound, Gas Chromatography Mass Spectrometry, and Isotope Ratio Analyses with Interpretable Machine Learning for Flavor Authentication: A Critical Analytical Review.

内容摘要

This critical review summarizes chiral volatile organic compound (VOC) analysis principles and compound-specific isotope ratio measurements, evaluating their combined application in flavor authentication. Natural flavorings are highly susceptible to adulteration, highlighting the need for authentication strategies that capture both molecular complexity and stereochemical specificity. We discuss recent advances in VOC profiling, emphasizing enantiomeric ratios (ERs) as biosynthetically grounded authenticity markers. Extraction strategies that preserve ERs are compared, and enantioselective gas chromatography-mass spectrometry (GC-MS) workflows are outlined, together with assessments of comprehensive two-dimensional GC (GC × GC) and heart-cut multidimensional GC (MDGC) for resolving co-elution. Quantitative and fingerprinting modes, library development, quality assurance protocols, and chemometric pipelines incorporating machine learning (ML) are evaluated for classification accuracy and interpretability. Orthogonal confirmation using isotope ratio mass spectrometry (IRMS) enhances decision confidence near ER thresholds. Applications across teas, fruits, essential oils, and juices show that reproducibility depends on standardized pretreatment, verified enantioselective separation, uncertainty-aware ER reporting, and leakage-safe model validation. Modular pipelines integrating GC × GC/MDGC, interpretable ML, and IRMS-with transparent, traceable outputs aligned to labeling regulations-offer a route toward industrial implementation. We also propose an integrated VOC/ER-centered framework combining chiral GC-MS, GC-combustion/pyrolysis-IRMS, and interpretable ML, emphasizing ER fidelity, and grey-zone decision rules for authentication.