Microsphere-enhanced flexible triboelectric nanogenerator with PVDF/CNT–PDMS composite for micro-expression recognition and mental health monitoring

Mental health disorders, especially those involving subtle somatization symptoms, urgently require continuous and noninvasive monitoring tools capable of capturing fine emotional cues. Microexpressions originate from transient, low-amplitude muscle activations and therefore demand sensing systems with high sensitivity and robust spatiotemporal resolution. In this work, we introduce a triboelectric sensing platform enabled by interface micro-engineering, realized through a microsphere-enhanced PVDF/CNT–PDMS composite architecture. By embedding uniformly distributed microspheres at the PDMS interface and coupling them with a PVDF/CNT electrospun layer, the engineered micro-contact configuration significantly increases effective contact area, local deformation uniformity, and interfacial charge density. The composite architecture maintains excellent flexibility and skin conformity, enabling stable performance under natural facial deformation. This design yields a sensitivity of 12.376 V·N⁻¹ and ensures stable electrical output under rapid and complex facial dynamics. When applied to key facial regions, the sensor resolves distinct pressure signatures associated with microexpression-level muscle fluctuations. Integrated with a Temporal Convolutional Network, the system recognizes seven emotional states with an accuracy of 93.43 % by extracting characteristic spatiotemporal features from both subtle and large-scale facial motions. This interface-engineered, self-powered sensing framework provides a promising route toward real-time and privacy-preserving mental health monitoring as well as more intuitive human–machine interaction.