Towards unobtrusive stress detection

摘要: This paper presents a machine-learning method for unobtrusive stress detection system. The idea is to automatically monitor three components of the stress response, ie emotional, behavioural and physiological. The emotional response is monitored using smartphone’s microphone and emotional voice analysis. The physiological response is monitored using a sensor-equipped wristband and bio-signal analysis. And, the behavioural response is monitored using smartphone’s sensors and smartphone-usage analysis. All three modules are combined using machine-learning methods to predict stress levels. We present approaches to monitor three components of the stress response with preliminary results and an indication of the feasibility of the proposed system.