# Evaluation of Astringent Compounds Using Electronic Tongue Technology

## TL;DR

The study shows that an electronic tongue can reliably assess the astringency of various compounds, matching human taste evaluations with high accuracy.

## Contribution

The paper introduces an integrated predictive model combining electronic tongue data and sensory evaluation for accurate astringency quantification.

## Key findings

- EGCG and EGC showed positive correlations between E-tongue responses and concentration.
- Procyanidin and EGCG exhibited the best linearity in correlation with sensory scores.
- Predictive models achieved high accuracy (R² > 0.9) in validating astringency assessments.

## Abstract

The electronic tongue (E‐tongue) is an emerging technology that enables the rapid and objective evaluation of various astringent compounds. This study investigated E‐tongue measurements and human sensory evaluation to analyze the concentration‐dependent astringency responses of seven compounds. Difference analysis, principal component analysis (PCA), and cluster analysis were applied to differentiate the compounds based on their taste profiles. The results revealed distinct relationships between E‐tongue response values and concentration. Specifically, epigallocatechin gallate (EGCG) and epigallocatechin (EGC) exhibited positive correlations with concentration, whereas tea polyphenols, tannic acid, and procyanidin showed negative correlations. In contrast, gallic acid and chlorogenic acid produced weak astringency responses. Difference analysis demonstrated significant taste variations across concentrations; PCA and cluster analysis further validated the E‐tongue's capability for distinct discrimination of astringent compounds. Regression analysis between E‐tongue measurements and human sensory scores demonstrated strong correlations for EGCG, EGC, tea polyphenols, tannic acid, and procyanidin, with procyanidin and EGCG exhibiting the best linearity. Predictive models achieved high accuracy (R
2 > 0.9, RMSE < 10%) in validation, demonstrating the E‐tongue's reliability as an alternative to sensory evaluation for astringency assessment.

This study investigated the capability of the electronic tongue to evaluate diverse astringent compounds. The integrated predictive model combining sensory evaluation and E‐tongue data accurately quantified astringency intensity, while multivariate analysis enabled clear discrimination and classification among different types of astringent compounds by the electronic tongue.