Research On The Key Techniques Of Psychological Assessment Based On Multi-source Information Fusion

The rapid development of modern information science and technology,such as smart wearable sensing technology,virtual reality(VR)technology,computer network technology,data mining technology,artificial intelligence technology,enables the research paradigm of psychology to be integrated with information science,which can effectively improve the traditional research methods and tools of psychometrics,and has broad application prospects and practical significance.Based on the cross-integration of psychology and information science,this paper studies the key technologies of of psychological assessment based on multi-source information fusion.The main research work and results of this paper includes:(1)In this paper,a universal intelligent psychological assessment system is designed and implemented.On the basis of investigating the current situation of psychometric work and informationization in different institutions such as hospital medical treatment,special education and prison management,this paper designs and implements in detail from the aspects of demand analysis,overall architecture,technical architecture,user-side software and hardware,administrator-side software,etc.The application layer of the overall architecture deploys user services and management services in Pod,implements micro-service management with Kubernates,and makes full use of the advantages of micro-service in distributed systems.Based on C/S architecture and using Microsoft.NET Framework 4.0 + Hybrid App to provide abundant psychological assessment content,the user-side designed and manufactured a humanized hardware platform,and reasonably configured three types of sensor acquisition devices:interaction,physiology and behavior.The administrator-side is based on the B/S architecture,and uses Nginx + Spring Cloud + Springboot + MyBatis+ Shiro + MySQL + Redis to build a distributed backstage management system to ensure the system runs reliably.The design and implementation of the system not only enhances the objectivity and effectiveness of psychological assessment,but also saves the human resources and time cost of the surveyor.It has certain universality for psychological measurement work in different fields and institutions.(2)In this paper,a P-value evaluation method of emotion based on feature-level fusion,an emotion recognition method based on decision-level fusion and a severe mental disease dia-nosis model based on multi-view model decision fusion are proposed.This paper comprehensively compares the advantages and disadvantages of three multi-source information fusion strategies:data-level fusion,feature-level fusion and decision-level fusion.In view of the superiority of feature-level fusion and decision-level fusion in the field of emotion recognition,firstly,this paper proposes a P-value evaluation method of emotion based on feature-level fusion.This method effectively solves the problem of multi-source information fusion at feature-level.It is fully applicable to the universal intelligent psychological assessment system proposed in this paper.Secondly,this paper proposes an emotion psychology state recognition method based on decision-level fusion,which combines expression-based emotion classifier and speech-based emotion classifier to recognize six basic emotion.This method is suitable for the universal intelligent psyychological assessment system proposed in this paper.Furthermore,the concept of multi-view is introduced into the diagnosis method of severe mental illness based on decision-level fusion,we synthesize the differences in image data,emotionally induced facial expression.behavior and physiological data,and the related gene expression data between people with mental health and people with severe mental illness.propose a severe mental disease diagnosis model based on multi-view model decision fusion.which can more comprehensively and accurately assess the mental health status of patients with severe mental illness.(3)Based on VR and intelligent scale,a series of studies are carried out to assess autistic children.Generally speaking,the assessment of ability deficits of autistic groups also belongs to the category of psychometrics.This paper studies the classical assessment scales for autistic children such as ABC?ADOS-G,ADI-R and ATEC.In view of the current situation of the assessment scales for autistic children in China,a preliminary assessment scale of Autism Assessment and Treatment Rating Scale for VR(ASTRS),which is localized in China and suitable for the realization of VR,is compiled.Secondly,we simplify the ALSOLIFE platform as a VR-based assessment-training system framework,and take the sub-scenarios in the framework as an example,innovatively propose an intelligent assessment method based on VR and the ASTRS.Finally,a VR-based social skills training system for autistic children is designed and implemented.A structured JA training method based on LTM prompt hierarchy and a method of attention distribution detection are proposed.The expected effectiveness of the system in assessing the social ability deficits of autistic children is demonstrated by experimental design and evaluation indicators.