Construction and application of machine learning models to screen functional ingredients for alcohol metabolism and its mechanism research

Zhu Tao, Yi Zheng, Xiaodi Jin, Xin Zhang, Xie He, Haitao Liu, Chunhua Fu, Longjiang Yu,
Construction and application of machine learning models to screen functional ingredients for alcohol metabolism and its mechanism research,
Food Bioscience,
Volume 74,
2025,
107685,
ISSN 2212-4292,
https://doi.org/10.1016/j.fbio.2025.107685.
(https://www.sciencedirect.com/science/article/pii/S2212429225018620)
Abstract: In this paper, we present the construction and application of machine learning models to screen functional ingredients that facilitate alcohol metabolism. These models evaluate and score compounds based on their structures to predict their potential to accelerate alcohol metabolism. The model parameters, including accuracy, precision, and F1 score, indicate that the constructed machine learning models exhibit strong performance. Notably, the models indicates that sterols, including glutenols and soya sterols, may have enhanced capabilities for accelerating alcohol metabolism. Consequently, we selected β-sitosterol, which exhibited the highest integrated probability, to design further experiments aimed at verifying its potential to mitigate the effects of alcohol. In alcohol-induced cell apoptosis model, a notable increase in cellular activity was observed when β-sitosterol was administered at a concentration of 50 μg/ml or higher prior to the addition of 200 mM alcohol. β-Sitosterol was observed to modulate ethanol dehydrogenase activity and reduce blood alcohol concentrations in an acute alcohol consumption rat model. Transcriptome analysis indicates that the administration of β-sitosterol significantly reduced the high expression of genes associated with hepatic Phase II reactions, such as Ugt1a5, and Gstp3, which were induced by chronic alcohol intake. Our experiments demonstrate that it is feasible to identify compounds present in food products that serve specific functions using AI technology. β-Sitosterol, which is recommended by a machine learning model, attenuates acute and chronic alcohol-induced damage by modulating ethanol dehydrogenase activity and hepatic Phase II reactions.
Keywords: β-sitosterol; Artificial intelligence; Alcohol metabolism; Alcohol dehydrogenase; UGT; GST