Sleep Planning with Awari: Uncovering the Materiality of Body Rhythms using Research through Design

Document Title

Sleep Planning with Awari: Uncovering the Materiality of Body Rhythms using Research through Design

Document Information

• Subject Area: Human-Computer Interaction and Health Informatics
• Keywords: Sleep, Self-Tracking, Health Informatics, Personal Informatics, Speculative Design, Research through Design

Research Background and Problem

• Identified Problems or Challenges:

  • With the growing prevalence of body tracking technologies, users can collect vast amounts of biometric data, but designers face challenges in transforming this data into comprehensible or actionable insights.
  • Current sleep tracking tools overly focus on passive data collection and linear goals, neglecting the complexity of body rhythms and the potential for planning future behaviors.
  • For "non-traditional sleepers," such as shift workers or polyphasic sleepers, current technologies fail to support their unique sleep patterns.

• Significance:

  • A deeper understanding of sleep rhythms and improved design can not only help users manage sleep but also enhance behavioral changes in other domains, such as diet and exercise.

• Research Motivation and Related Work:

  • Using research through design to explore the material characteristics of sleep rhythms and how design can influence body rhythms.
  • Collaborating with participants to reimagine sleep tracking as an active tool for planning future wakefulness rather than a passive data collection device.

Solution

• Proposed Method or Solution:

  • Developed a web-based interface called Awari, which uses personal sleep data and reinforcement learning algorithms to predict users' wakefulness and assist in sleep planning.
  • During the design process, conducted workshops with "non-traditional sleepers" (e.g., night-shift nurses, polyphasic sleepers) to gather feedback and guide model iterations.
  • Proposed three non-exclusive categories of sleep rhythms: slow and cyclic rhythms, stress and release rhythms, and anchoring rhythms, designing the system to present and manage these rhythms.

• Innovations:

  • Shifted from passive sleep data tracking to predicting and planning future wakefulness trajectories.
  • Introduced the material characteristics of body rhythms into the design, such as rhythm periodicity, the impact of stress on release, and the role of anchoring actions.
  • Enabled users to explore and plan rhythms by adjusting parameters (e.g., light preferences, stress tolerance).

• Implementation Steps and Key Technologies:

  • Developed the Awari system through four iterations:
    1. Static Model: Allowed users to freely set their circadian rhythms and explore how past sleep behaviors influenced wakefulness.
    2. Random Model: Used random exploration of users' future sleep patterns to predict wakefulness.
    3. Reinforcement Learning Model: Leveraged a multi-objective reinforcement learning algorithm to balance users' sleep, wakefulness, and social responsibilities.
    4. Anchoring Model: Incorporated sleep inertia and the influence of light on rhythms to address unreasonable micro-sleep suggestions.

Research Outcomes

• Specific Outcomes:

  • The developed Awari system enables users to explore, plan, and adjust their future sleep and wakefulness states.
  • Identified three categories of sleep rhythms and their associated material characteristics, offering new perspectives for designing behavior change technologies.
  • Proposed an interactive medical model that allows users to directly participate in adjusting rhythm models rather than passively relying on system recommendations.

• Advantages Over Existing Solutions:

  • The system emphasizes user agency and autonomy, providing an interactive space for users to reflect on and plan their sleep.
  • The design supports more flexible sleep patterns (e.g., segmented sleep, polyphasic sleep), aligning better with real-world needs.
  • Shifted from passive data collection to active planning, enhancing users' understanding of complex body rhythms.

• Experimental or Evaluation Results:

  • Feedback from workshop participants highlighted the necessity of dynamic planning and prioritizing social responsibilities.
  • The system efficiently plans complex sleep patterns, aligning closely with real user behaviors.

• Limitations and Future Directions:

  • Limitations: The current design has not been thoroughly tested for long-term user experience, and simulated data does not fully validate the complexity of real-world data.
  • Future Directions:
    • Further explore the application of multi-objective reinforcement learning models in wakefulness planning.
    • Investigate the effects and behavioral changes of long-term use of this technology in real-life scenarios.
    • Develop additional design models to support other body rhythms (e.g., dietary, exercise rhythms).

Discussion and Contributions

• The Future of Self-Tracking Technology: Providing users with tools to explore and plan body rhythms expands the scope of interaction design from focusing on past behaviors to involving future actions, enhancing user agency.
• Democratization of Medical Models: The system enables users to intuitively understand how medical algorithms affect their rhythms, particularly supporting "non-standard users," contributing to fairness and inclusivity in behavior change.
• Technology-Enhanced Life Rhythms: Design should consider the combined impact of social, chemical/physical, and psychological factors on rhythms, offering new possibilities for performance optimization and behavior change design.