Dongfu Xu

Mastication and swallowing are complex physiological processes involving the coordinated activity of multiple tissues in the oral cavity, facial region, and laryngeal system. Some detection methods suffer from limitations such as insufficient information acquisition and inadequate temporal feature analysis. To address these issues, this study proposes a conceptual method for analyzing the state of masticatory and swallowing movements. It integrates maxillofacial electromyographic (EMG) signals with laryngeal movement signals. The goal is to preliminarily explore state analysis of masticatory and swallowing movements over time. A designed gain-adjustable conditioning circuit processes and acquires these signals: maxillofacial EMG signals from EMG electrodes and laryngeal movement signals from flexible PVDF piezoelectric sensors. These two signal streams complement each other’s missing information, enabling comprehensive detection of the state of masticatory and swallowing movements. To address time-point labeling in mastication and swallowing, a sliding-window-based dispersion calculation method was employed to extract characteristic signal nodes, which were then accurately associated with their corresponding physiological motion states. We combined temporal features such as the zero point, onset of fluctuations, characteristic peaks, and baseline recovery from electromyographic (EMG) signals and laryngeal movement signals. This allowed us to establish a correspondence between key time points in the mastication and swallowing processes. The coefficient of determination (R²) for the pressure–voltage linear fit of the PVDF flexible piezoelectric sensor was 0.99446. The pressure resolution was approximately 0.08 kPa. Response times were no more than 15 ms for the EMG channel and no more than 10 ms for the PVDF pressure channel. These results indicate that this method is feasible for extracting oral movement time parameters in healthy subjects.

End-tidal breath gases originate deep within the lungs, and their composition is an especially accurate reflection of the body’s metabolism and health status. Therefore, accurate collection of end-tidal breath gases is crucial to enhance electronic noses’ performance in breath detection. Regarding this issue, this study proposes a novel electronic nose system and employs a threshold control method based on exhaled gas flow characteristics to design a gas collection module. The module monitors real-time gas flow with a flow meter and integrates solenoid valves to regulate the gas path, enabling automatic collection of end-tidal breath gas. In this way, the design reduces dead space gas contamination and the impact of individual breathing pattern differences. The sensor array is designed to detect the collected gas, and the response chamber is optimized to improve the detection stability. At the same time, the control module realizes automation of the experiment process, including control of the gas path state, signal transmission, and data storage. Finally, the system is used for breath detection. We employ classical machine learning algorithms to classify breath samples from different health conditions with a classification accuracy of more than 90%, which is better than the accuracy achieved in other studies of this type. This is due to the improved quality of the gas we extracted, demonstrating the superiority of our proposed electronic nose system.