Application Research Of Multi-perception Wearable Device In Mental Health Monitoring

Long-term wellbeing monitoring is an underlying theme in many local and national policies and procedures outlined by governments and health care services.Naturally,efficacious,and trustworthy monitoring by using wearable sensors is necessary for researchers to find and establish the interrelationships of Affective Computing(AC),Body Sensor Networks(BSNs),Social Signal Processing(SSP)and Physical Mental Health(PMH).Specially,Giancarlo Fortino,et al.have an outstanding contribution for applying BSNs on health monitoring.This paper investigated how technology can help to objectively monitor an individual’s wellbeing in a naturalistic environment.For this purpose,we designed and implemented a wearable device with the integration of multisensors which consist of audio sensing,behavior monitoring,environment and physiological sensing.In order to avoid privacy issues,four audio-wellbeing features are embedded into a wearable hardware platform to automatically evaluate speech information without preserving raw audio data.In addition,four weeks of long-term monitoring experiment studies have been conducted in conjunction with a series of wellbeing questionnaires in a group of students.The relationships between physical and mental health were investigated objectively by utilizing data from speech,behavioral activities and ambient factors in a completely natural daily situation.The main research contents of this paper are as follows:1)Based on the multi-sensor wearable technology,this paper designs and implements a wearable watch device to collect multi-sensors data.2)Design and implement a long-term mental health monitoring experiment,this parper conducts a four-week ASQ tendency experiment and collects nearly 200 GB data and more than 300 mental health questionnaires.3)For the lowfrequency and audio data,this parper proposes a feature analysis method.4)According to the correlation between audio features,behavioral features and autism scores,this parper have compared and analyzed the differences of ASQ high-scorers and ASQ lowscorers.5)An evaluation model of college students’ autism tendency based on Boosting algorithm using multi-sensor features is proposed.