Logo Image
News|Articles|April 10, 2026

HS-SPME/GC–MS and Machine Learning Enable Volatile Fingerprinting and Classification of Commercial Vinegars

Author(s)John Chasse
Listen
0:00 / 0:00

Key Takeaways

  • HS‑SPME/GC‑MS enabled broad volatile coverage across four vinegar matrices, establishing an analytical basis for cross-type differentiation beyond process-condition studies.
  • Comparative modeling showed Random Forest outperformed XGBoost, KNN, SVM, and MLP for multiclass vinegar identification, achieving 96.19% accuracy with stable generalization within the dataset.
SHOW MORE

Headspace solid-phase microextraction coupled with GC–MS (HS-SPME/GC-MS), integrated with machine learning, enabled comprehensive profiling of 127 volatile compounds across multiple vinegar types. Random Forest modeling achieved high classification accuracy, identifying key aroma markers and demonstrating the power of chromatographic–data science workflows for differentiating fermented food products.

Although the aroma diversity of vinegar arises from volatile metabolites formed during the product’s alcoholic and acetic acid fermentation, analyses of volatile profiles across types of vinegar remain limited. In response, researchers characterized four vinegar types (rice, grape, apple, and persimmon) using headspace solid-phase microextraction coupled with gas chromatography-mass spectrometry (HS-SPME/GC-MS) and machine learning. A paper based on this work was published in Food Chemistry.1

A traditional fermented food produced through sequential alcoholic and acetic acid fermentations of fruit- or grain-based substrates, vinegar’s fermentation processes create a wide range of volatile compounds that all play a part in developing its characteristic flavor and aroma.2 These compounds play an important part in the determination of the overall sensory quality and consumer preference because they reflect the chemical traits of the raw materials as well as the metabolic activities of the microorganisms involved in the vinegar’s fermentation.3

Despite the long history of vinegar, the number comparative investigations that analytically assess and categorize volatile compositions across different vinegar types is considered inadequate, as previous studies have mainly been concentrated on the influence of fermentation conditions, microbial strains, duration of aging, or the methods of processing on the development of aroma. 4-8

The research team’s analysis identified127 volatiles from 87 commercial samples of the quartet of vinegar types. To ensure methodological consistency with previous studies and comprehensively evaluate different algorithmic principles, the team employed five machine learning algorithms: Extreme Gradient Boosting (XGBoost), K-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Machine (SVM), and Multilayer Perceptron (MLP). Random Forest achieved the most stable performance (96.19% accuracy). The Random Forest with Recursive Feature Elimination and Cross-Validation (RF-RFECV) model selected 21 predictive features while 18 corresponded to known fermentation-derived volatiles; three were identified as non-intentionally added substances or packaging-related contaminants. Shapley Additive exPlanations (a game-theory-based, model-agnostic method used to interpret machine learning model predictions) revealed type-specific drivers, including methyl acetate and 2-acetoxy-3-butanone (persimmon), acetic acid and isobutanol (grape), and benzaldehyde (rice).1

“Overall,” write the authors of the paper,1 “this study presents an exploratory, data-driven framework for volatile-based vinegar differentiation. The findings are inherently dataset-specific and warrant further validation using independent sample set to assess their broader applicability.”

The research team is of the opinion that future studies incorporating larger cohorts and external validation will be essential to substantiate potential industrial applications.1

References

  1. Park, S.; Kim, K.; Sung, J. et al. Classification of Vinegar Types Using Volatile Compound Profiles and Machine Learning. Food Chem. 2026, 514, 149076. DOI: 10.1016/j.foodchem.2026.149076
  2. Ge, Y.; Wu, Y.; Aihaiti, A. et al. The Metabolic Pathways of Yeast and Acetic Acid Bacteria During Fruit Vinegar Fermentation and Their Influence on Flavor Development. Microorganisms 2025, 13 (3), 477. DOI: 10.3390/microorganisms13030477
  3. Shi, H.; Zhou, X.; Yao, Y. et al. Insights into the Microbiota and Driving Forces to Control the Quality of Vinegar. LWT2022, 157, 113085, DOI: 10.1016/j.lwt.2022.113085
  4. Wang, W.; Zhang, F.; Dai, X. et al. Changes in Vinegar Quality and Microbial Dynamics During Fermentation Using a Self-Designed Drum-Type Bioreactor. Front Nutr. 2023, 10, 1126562. DOI: 10.3389/fnut.2023.1126562
  5. Zhang, L.; Wang, M.; Song, H. et al. Changes of Microbial Communities and Metabolites in the Fermentation of Persimmon Vinegar by Bioaugmentation Fermentation. Food Microbiol. 2024, 122, 104565. DOI: 10.1016/j.fm.2024.104565
  6. Al-Dalali, S.; Zheng, F.; Sun, B. et al. Tracking Volatile Flavor Changes During Two Years of Aging of Chinese Vinegar by HS-SPME-GC-MS and GC-O. J. Food Comp. Anal. 2022, 106, 104295. DOI: 10.1016/j.jfca.2021.104295
  7. Ji, X.; Xu, L. Comparison and Analysis of the Volatile Compounds in Solid-State and Liquid-State Fermented Vinegars. J. Food Meas. Charact. 2022, 16, 4914-4922. DOI: 10.1007/s11694-022-01590-0
  8. M. Wagner, M.; Zaldarriaga Heredia, J.; Segura-Borrego, M. P. et al.

Identification of Potential Volatile Markers for Characterizing Argentine Wine Vinegars Based on their Production Process. Talanta Open2024, 10, 100370. DOI: 10.1016/j.talo.2024.100370

Related Content