Machine learning-enhanced flavoromics: Identifying key aroma compounds and predicting sensory quality in sauce-flavor baijiu.

The quality of Sauce-flavor baijiu hinges on sensory characteristics and key aroma compounds, which traditional methods struggle to evaluate accurately and effectively. This study explores the sensory characteristics and aroma compounds of Sauce-flavor baijiu across different rounds using flavoromics and machine learning, constructing quality grade prediction models. Sensory characteristics shift from acid in the early stages BJ1-BJ2 rounds to sauce in the mid-stages BJ3-BJ5 rounds and caramel in the late stages BJ6-BJ7 rounds. Employing AEDA and OAV analyses, 18 key odor-active compounds were identified, such as ethyl butyrate, ehyl isovalerate, and phenethyl acetate. Additionally machine learning models combined with clustering algorithms achieved high accuracy in predicting quality grades: 85 % (MLP+ HCA), 97 % (XGBoost+ K-means), and 84 % (Random Forest+ GMM). The SHAP model identified 20 key aroma compounds, including diethyl succinate, Tetramethylpyrazine, and Acetaldehyde, determining quality concentration thresholds. This study offers robust methods for baijiu flavor control and quality evaluation.
Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
(Copyright © 2025 Elsevier Ltd. All rights reserved.)