Touch Sense

Non-intrusive and sensitive measurement of users' stress is very crucial for computers to dynamically understand and respond to users' mental status while users are naturally interacting with them, such as context-aware reminding, smart assistant, health monitoring etc. Compared to others, physiological measures are known as more sensitive and reliable ones. However, most of the current physiological methods need explicit and obtrusive sensors that deter natural human-computer interactions. In this study, we propose a photoplethysmogram-based mental stress measuring method through infrared touchscreen, which, from the best of our knowledge, is the first to integrate with the touch modality to enable natural, sensitive and reliable measurement. We designed and conducted two user experiments with touch-and-hold mode and tap mode respectively. By using person-independent classifiers to handle the photoplethysmographic parameters, the results reached a 97% and 87% average stress recognition accuracy for static test and interaction test respectively.