Exploring the use of face video to measure mental health
LE3 .A278 2021
2021
Leslie, Kenneth
Acadia University
Bachelor of Science
Honours
Psychology
Covid-19 frontline workers are at risk for depression, moral injury, and suicide. People who need help are often reluctant to seek help, and so a mechanism for detecting emotional distress from face video could be part of a more comprehensive plan to detect and treat mental distress in frontline workers. Here we explore the potential application of face video analysis to detect depression CESD-R), attachment style (ECR-RS), mood (POMS-SF), and empathy (EQ) in an undergraduate student population. Face video was collected remotely using iMotions software. Facial expressions and Eye Blink Rate (EBR) were measured using the AFFDEX automated facial affect coding algorithm and an R script. Facial responses were measured while watching a video, speaking a narrative, and performing a range of face imitation tasks. The face imitation tasks consisted of making emotional faces on command (COMMAND), imitating videos of facial expressions from one of four models (IMITATE), and making the opposite expression (OPPOSITE). These tasks were repeated in two conditions: a POSITIVE VALENCE condition, featuring a positive video and narratives in response to questions about YouTube, and a NEGATIVE VALENCE condition, featuring a Covid-19 news video and narratives in response to questions about Covid-19. For the preliminary analysis, we looked at expression of joy vs. fear across all conditions, as well as EBR. There were more expressions, p = .016, and micro expressions, p = .038, of joy during the positive videos. In all of the motor tasks, AFFDEX was much stronger at detecting expressions of joy, p< .001. EBR was lower during the surveys than during both narratives and the negative video, p < .001.
The author retains copyright in this thesis. Any substantial copying or any other actions that exceed fair dealing or other exceptions in the Copyright Act require the permission of the author.
https://scholar.acadiau.ca/islandora/object/theses:3568
