Stress Management Training using Biofeedback guided by Social Agents

Title of the Paper

Stress Management Training using Biofeedback guided by Social Agents

Paper Information

• Authors: Tanja Schneeberger, Naomi Sauerwein, Manuel S. Anglet, Patrick Gebhard
• Field: Human-Computer Interaction, Applied Psychology, Health Technology
• Keywords: Social Agents, Stress Management Training, Biofeedback, Mental Health, Human-Computer Interaction
• Conference: 26th International Conference on Intelligent User Interfaces (IUI 2021)
• Link: https://doi.org/10.1145/3397481.3450683

Research Background and Problem

• Problems and Challenges:
  • Prolonged psychological stress poses severe risks to physical and mental health, including depression and cardiovascular diseases.
  • Existing biofeedback training requires the presence of a coach, with limited technical support during the process. Moreover, traditional stress management methods (e.g., stress diaries) have limited effectiveness in raising awareness.
• Research Importance:
  • The World Health Organization identifies stress as one of the most significant health risks in modern society.
  • Effective stress management methods are critical to preventing sick leave or productivity losses caused by mental health issues.
• Research Motivation:
  • By integrating advanced social agents in human-computer interaction (e.g., providing emotional simulation and non-verbal behavior) with biofeedback technology, the study aims to develop a novel stress management approach.
  • The goal is to enhance participants' ability to cope with real social stress and prevent health risks associated with chronic stress.

Solution

• Method:
  • A virtual stress management training system was developed, combining heart rate variability (HRV)-based biofeedback technology with interactive social agents.
  • The social agent "Gloria" was used to provide guidance, replacing traditional human biofeedback coaches.
  • The training process was divided into two phases:
    1. Visualization Feedback Phase: Combining biofeedback monitors with advice from Gloria.
    2. Verbal Feedback Phase by Gloria Only: Gradually removing technical support to increase the sense of social context.
• Innovations:
  • For the first time, a virtual social agent completely replaced human coaches in biofeedback training.
  • Real or simulated social stressors (e.g., image stimuli, task challenges) were integrated into the training to promote the transfer of learning to real-life situations.
• Key Technologies:
  • Plux wireless biosignal acquisition tools were used to collect heart rate and respiration rate data.
  • A real-time multi-component system was implemented, including signal interpretation, interaction management, and biofeedback monitoring.
  • The VisualSceneMaker (VSM) toolkit was applied to create Gloria's verbal and non-verbal behaviors.

Research Results

• Experimental Design:

  • 71 participants were divided into an experimental group (EG) and a control group (CG).
  • The experimental group underwent biofeedback training guided by Gloria, conducted twice; the control group used traditional stress diary methods.
  • Participants' stress levels and self-performance were measured before and after a stress-inducing task (Trier Social Stress Test, TSST).

• Key Findings:

  1. Stress Reduction:
     • The experimental group reported significantly lower stress levels after training and the stress-inducing task compared to the control group, indicating the superior effectiveness of biofeedback training in stress relief.
  2. Task Stressfulness:
     • The experimental group reported lower subjective task stressfulness, though the difference was not statistically significant.
  3. Self-Performance:
     • The experimental group scored significantly higher on self-performance evaluations compared to the control group.
  4. Physiological Data and Self-Assessment:
     • Lower stress levels and higher heart rate variability were associated with better task performance and stress management capabilities.

• Comparison with Existing Methods:

  • Compared to the stress diary method, biofeedback training guided by social agents demonstrated better stress reduction effects and improved participants' adaptation to social stress scenarios.

• User Experience:

  • Participants gave overall positive feedback, particularly praising the novelty of the system.
  • Expert evaluations also highlighted the system's high potential for stress prevention in therapeutic and workplace settings.

• Limitations and Future Directions:

  1. Training Duration: The experiment involved only two training sessions; further research is needed to examine the effects of increased training frequency.
  2. Comparison with Human Coaches: The study did not compare the effectiveness of virtual coaches with human coaches, which should be addressed in future research.
  3. Long-Term Effects: The study lacked long-term follow-up data on training effects; future studies could include follow-up assessments.
  4. Mobile Platform Expansion: Future goals include developing a mobile version of the biofeedback training system for smartphones or smartwatches.

Conclusion

This study validated the effectiveness of biofeedback training guided by social agents in stress management, filling a research gap in the field of virtual biofeedback coaching. It demonstrated the potential of this approach in clinical and preventive health applications.