A Mask-Shaped Respiration Sensor Using Triboelectricity and a Machine Learning Approach toward Smart Sleep Monitoring Systems

Abstract

Daily sleep monitoring is limited by the needs for specialized equipment and experts. This study combines a mask-shaped triboelectric nanogenerator (M-TENG) and machine learning for facile daily sleep monitoring without the specialized equipment or experts. The fabricated M-TENG demonstrates its excellent ability to detect respiration, even distinguishing oral and nasal breath. To increase the pressure sensitivity of the M-TENG, the reactive ion etching is conducted with different tilted angles. By investigating each surface morphology of the polytetrafluoroethylene films according to the reactive ion etching with different tilted angles, the tilted angle is optimized with the angle of 60° and the pressure sensitivity is increased by 5.8 times. The M-TENG can also detect changes in the angle of head and snoring. Various sleep stages can be classified by their distinctive electrical outputs, with the aid of a machine learning approach. As a result, a high averaged-classification accuracy of 87.17% is achieved for each sleep stage. Experimental results demonstrate that the proposed combination can be utilized to monitor the sleep stage in order to provide an aid for self-awareness of sleep disorders. Considering these results, the M-TENG and machine learning approach is expected to be utilized as a smart sleep monitoring system in near future.