A pervasive stress monitoring system based on biological signals

In this research, we focus on detecting stress based on electroencephalogram (EEG) method. An experiment has been conducted with 59 subjects, the results show that three EEG features from Fpz point, LZ-complexity, alpha relative power and the ratio of alpha power to beta power, are effective respectively in the stress detection using K-Nearest-Neighbor classifier, however Naive Bayesian classifier is not suitable for the stress prediction based EEG data. Meanwhile, we introduced the stress index for indicating stress level. Based on these work, we build a pervasive stress detection system which enables people to monitor their stress level opportunely. The proposed system provides services both for ordinary users in 'User Panel' and psychiatrists in 'Doctor Panel'. The 'User Panel' integrates biological signals acquisition which collects user's EEG data for stress classification, self-assessment questionnaire as reference to stress index, history record for logging user's state, and chatting with doctor, aiming to keep in touch with psychiatrists if necessary. In 'Doctor Panel', psychiatrists can view all users' historical status and chat with them.