Machine learning-driven investigation of key flavor factors in pale lager beer fermentation samples and their evolution mechanisms during fermentation.

The dynamic evolution of flavor compounds during industrial pale lager fermentation was poorly characterized, which hindered the achievement of precise quality control. In this study, flavoromics, sensory analysis, and machine learning were integrated to systematically map the dynamic changes of compounds and the evolution of sensory characteristics throughout the fermentation of pale lager. Malty aroma and hop aroma were identified as the core sensory attributes of the beer fermentation samples. SHAP analysis revealed the key drivers of these sensory attributes: 2-methylpyrazine and maltol for malty aroma, and methyl geranate and myrcene for hop aroma. Subsequently, these findings were validated by recombination and omission experiments. Based on these results, this study further explored the process parameters influencing the key flavor compounds through correlation analysis. This study elucidated the key flavor factors of pale lager, laying a foundation for the intelligent monitoring and precise regulation of flavors during the industrial brewing process.