Machine learning-based detection of acute psychosocial stress from body posture and movements.

Investigating acute stress responses is crucial to understanding the underlying mechanisms of stress. Current stress assessment methods include self-reports that can be biased and biomarkers that are often based on complex laboratory procedures. A promising additional modality for stress assessment might be the observation of body movements, which are affected by negative emotions and threatening situations. In this paper, we investigated the relationship between acute psychosocial stress induction and body posture and movements. We collected motion data from N = 59 individuals over two studies (Pilot Study: N = 20, Main Study: N = 39) using inertial measurement unit (IMU)-based motion capture suits. In both studies, individuals underwent the Trier Social Stress Test (TSST) and a stress-free control condition (friendly-TSST; f-TSST) in randomized order. Our results show that acute stress induction leads to a reproducible freezing behavior, characterized by less overall motion as well as more and longer periods of no movement. Based on these data, we trained machine learning pipelines to detect acute stress solely from movement information, achieving an accuracy of 75.0 ± 17.7 % (Pilot Study) and 73.4 ± 7.7 % (Main Study). This, for the first time, suggests that body posture and movements can be used to detect whether individuals are exposed to acute psychosocial stress. While more studies are needed to further validate our approach, we are convinced that motion information can be a valuable extension to the existing biomarkers and can help to obtain a more holistic picture of the human stress response. Our work is the first to systematically explore the use of full-body body posture and movement to gain novel insights into the human stress response and its effects on the body and mind.

Comparing Direct and Indirect Methods of Audio Quality Evaluation in Virtual Reality Scenes of Varying Complexity.

Many quality evaluation methods are used to assess uni-modal audio or video content without considering perceptual, cognitive, and interactive aspects present in virtual reality (VR) settings. Consequently, little is known regarding the repercussions of the employed evaluation method, content, and subject behavior on the quality ratings in VR. This mixed between- and within-subjects study uses four subjective audio quality evaluation methods (viz. multiple-stimulus with and without reference for direct scaling, and rank-order elimination and pairwise comparison for indirect scaling) to investigate the contributing factors present in multi-modal 6-DoF VR on quality ratings of real-time audio rendering. For each between-subjects employed method, two sets of conditions in five VR scenes were evaluated within-subjects. The conditions targeted relevant attributes for binaural audio reproduction using scenes with various amounts of user interactivity. Our results show all referenceless methods produce similar results using both condition sets. However, rank-order elimination proved to be the fastest method, required the least amount of repetitive motion, and yielded the highest discrimination between spatial conditions. Scene complexity was found to be a main effect within results, with behavioral and task load index results implying more complex scenes and interactive aspects of 6-DoF VR can impede quality judgments.