Analysis and Predictive Modeling of In Vitro Digestive Stability of Tea Polyphenols in 24 Tea Varieties

Abstract

Tea polyphenols (TP) constitute key functional components in tea leaves, yet their instability during human digestion limits absorption. This study selected 24 representative tea varieties to establish an in vitro dynamic digestion model, investigating the dynamic changes of TP during digestion. Characteristic signals of tea liquor were collected using electronic nose and color difference technology to construct predictive models, enabling the estimation of TP content (TPC) before and after digestion. Results indicate the tea liquor aroma is dominated by hydrogen compounds, and tea fermentation level correlates positively with liquor color intensity. Green tea exhibits the highest TPC and catechin levels, demonstrating potent antioxidant properties. TP remains relatively stable in the stomach but is significantly reduced in the intestine, whilst the catechin components exhibit distinct degradation patterns. The Extreme Gradient Boosting (XGBoost) model predicting TPC based on electronic nose data demonstrated the highest efficacy, with an R²: 0.92. The XGBoost model incorporating color difference and TPC achieved the best performance, with an R²: 0.81. This research provides robust support for elucidating TP digestion patterns in humans and establishing a tea quality assessment system.