A tongue-mimicking gel sensor with efficient feature extraction for mechanistic studies of astringency perception

In the study of taste perception, researchers aim to achieve the goal of simulating the human tongue’s sensory process through machines, and reproducing taste functions in vitro. Based on the principle of astringency perception, a tongue-mimicking gel sensor was designed based on the principle of astringency perception, replicating its composition, content, and cross-linked structure. This design enables the sensor’s internal signal variations to closely reproduce the key molecular dynamic processes of the salivary pellicle, thereby achieving in vitro simulation of polyphenol-induced astringency. However, conventional tongue-mimicking gel sensor analyses rely solely on conductivity signals, which are insufficiently comprehensive and insufficient to capture the complex dynamics of the sensing process. To address this limitation, in-depth research into the reaction kinetics within the tongue-mimicking gel sensor is essential for a more precise interpretation of oral astringency perception mechanisms. Consequently, developing feature processing methods capable of analyzing dynamic changes – such as molecular recognition and signal conversion within the system – has emerged as a critical scientific challenge requiring important resolution. Therefore, the present study employs electrochemical impedance spectroscopy (EIS) combined with signal transduction analysis within the tongue-mimicking gel sensor to establish an equivalent circuit model. This model enables stage-by-stage interpretation of the reaction kinetics through the variations in component parameters. Through systematic analysis of how these parameters change with astringent compound concentrations and investigation of their correlation with the sensor’s conductivity, the magnitude of the constant phase element parameter Q (CPE-Q) was extracted as the most significant feature. The variations in CPE-Q reflect its crucial role in both signal transduction and astringency sensing, making it a key parameter for characterizing the sensor’s dynamic response mechanism. This newly developed feature selection method for the tongue-mimicking gel sensor utilizes the CPE-Q parameter to comprehensively capture the dynamic information involved in astringency recognition. By effectively overcoming the informational limitations inherent in traditional conductivity-based measurements, this approach achieves a significant improvement in prediction accuracy. Furthermore, based on this advancement, the study establishes a theoretical model that enables systematic analysis of the molecular dynamics underlying astringency recognition within the sensor system. This model bridges biomimetic sensing signals with physiological taste mechanisms, thereby providing a novel platform and perspective for mechanistic research and application development in related fields.