Wearable Social Perception Physical And Mental Health Assessment System

Social signal processing and affective computing have been extensively studied in various fields such as medical,psychology,and sociology.Studies have shown that mining social information can infer complex human social behaviors through the fusion of different data features.With the improvement of people's material living standards,more and more people have begun to focus on mental health issues.The traditional method of mental health diagnosis is to obtain clinical data though observation,interview,experiment and questionnaire,and to judge the psychological status and personality characteristics of the diagnosed person through the analysis of the data.The traditional assessment method is inevitably interfered by the subjective and empirical factors of the diagnostician,which affects the accuracy of the assessment.Therefore,this paper designs a wearable social perception system to collect the wearer's speech and behavior data during the speech process,extracts the social speech and social behavior features,and establishes the anxiety status evaluation model of college students by analyzing the Pearson correlation between social features and anxiety scale scores.The main research contents of this dissertation are as follows: 1)Based on social perception technology,this paper designs a type of wearable device which can extract social features and objectively assess mental health state.2)Based on the uniqueness of the functionality of wearable devices and the popularity of smart phones,this paper designs a social-aware App and an Alibaba Cloud server system.3)This paper proposes an analysis method for social features.It includes 76 kinds of speech features and 43 kinds of behavioral features.By organizing college students to perform anxiety experiments,this paper studies the correlation between social features and anxiety state.4)On the premise of feature selection and feature compression,the correctness and computational complexity are compared between SVM classifiers with linear kernel functions,SVM classifiers with RBF kernel functions,and Logistic Regression(LR)classifiers to determine the use of the LR classifier for pre-classification.5)This paper proposes an anxiety assessment model for college students based on social perception features.