Detection of Psychological States from Visual Cues in University Students

Abstract

Psychological health or mental health is fundamental to human health and is the most newlineimportant aspect of human being. Depression, anxiety, and stress are psychological states that newlinecan disrupt mental stability. Often these disorders go undetected in young people and later on newlinelead to many other comorbidities directly or indirectly. The increasing rate of depression, newlineanxiety, and stress among university students is a matter of great concern worldwide. Early newlinedetection of such mental health issues can be highly beneficial for the individual who seeks newlinehelp and also for mental health practitioners. However, lack of trained mental health care newlineproviders, inconsistent and ambiguous outcomes of self-report inventories that many a time newlinebypass clinical significance, and subjective bias associated with clinician-administered rating newlinescales are the major bottlenecks in early detection. With the advances in computer vision newlinetechnology, visual cue-based non-invasive methods are becoming increasingly popular for the newlineassessment of mental health disorders. Among the visual cues, facial expressions, eye newlinemovement features, and body postures have been studied widely by researchers. Inspired by newlinenon-invasiveness, simplicity of approach, and richness of eye movement features, this thesis newlineexamines the feasibility of using eye movement features for assessment of each of newlinedepression, anxiety, and depression with the support of a video-based emotional elicitation newlinesetup. Two important eye movement features, i.e., blink patterns and pupil dilation patterns newlineunder different elicited emotions, are studied. Further, these features are subjected to newlinestatistical and machine learning (ML) analyses to verify their appropriateness. newlineThe objectives of this thesis are to investigate the usefulness of eye blink features and newlinepupil dilation features in the diagnosis of depression, anxiety, and stress. As a preliminary newlinestep, a three-stage design framework is proposed and an experimental protocol is defined to newlinecollect data. The data contains recorded face videos of

Description

Keywords

Citation

item.page.endorsement

item.page.review

item.page.supplemented

item.page.referenced