Co-designing a virtual reality based mindfulness application to address diabetes distress using Artificial Intelligence-informed Experience-Based Co-Design (AI-EBCD): a feasibility study

This feasibility study utilized Artificial Intelligence-informed Experience-Based Co-Design (AI-EBCD) to gather insights from adults with type 2 diabetes and mindfulness practitioners, resulting in a tailored prototype for a virtual reality mindfulness application aimed at alleviating diabetes distress.

The Gist

Imagine you are trying to learn how to swim, but the pool is crowded, the water is cold, and you're constantly reminded of how much you're struggling to stay afloat. This is what it feels like for many people living with Type 2 Diabetes. They aren't just managing a physical condition; they are dealing with "diabetes distress"—a heavy emotional burden of worry, frustration, and feeling overwhelmed by the daily demands of checking blood sugar, eating right, and taking medication.

This research paper is about building a digital life raft to help these people find calm. But instead of just a regular app, the researchers wanted to build a Virtual Reality (VR) life raft—a place where you can put on a headset and step into a completely different world to practice mindfulness (the art of staying calm and present).

Here is the story of how they built it, explained simply:

The Problem: The "One-Size-Fits-All" Life Jacket Doesn't Fit

The researchers knew that mindfulness is great for stress, but existing solutions had holes in them:

• Face-to-face classes are like expensive private swimming lessons; they are great but hard to get to and pay for.
• Standard Apps are like reading a swimming manual while standing on the beach. They are accessible, but they lack the "immersion" to really help you forget your worries.
• Existing VR Apps were like life jackets designed for the ocean, not for a specific type of swimmer. No one had asked the people with diabetes what they actually wanted in a VR mindfulness tool.

The Solution: Co-Designing the "Perfect Life Jacket"

Instead of the scientists sitting in a lab and guessing what people needed, they decided to co-design the solution. Think of this as inviting the swimmers into the workshop to help sew the life jacket.

They used a special recipe they called AI-EBCD (Artificial Intelligence-informed Experience-Based Co-Design). Here's how the process worked, step-by-step:

1. Listening to the Experts (The Lifeguards):
   First, they interviewed 9 mindfulness experts. These were the "lifeguards" who knew the rules of the water. They told the team: "Keep it simple. Don't make the interface complicated, or people will panic. Make sure there's a safety button if someone gets scared."

2. The Workshop with the Swimmers (People with Diabetes):
   They gathered 13 adults with Type 2 diabetes for a series of creative workshops.

   • The Art Phase: Participants drew their ideal calm place. Some wanted forests, others wanted beaches.
   • The AI Magic Phase: This was the cool part. They used Generative AI (like a super-smart robot artist). The participants typed prompts like "a peaceful forest with soft blue light," and the AI instantly created images and sounds. This let people "see" their ideas come to life immediately, rather than just describing them.
   • The Voting Phase: They tried out different features and voted on what mattered most.

What Did They Discover? (The Blueprint)

The "swimmers" and "lifeguards" agreed on a few key things for their perfect VR life jacket:

• Keep it Simple (The "No-Clutter" Rule):
  One participant said, "If the menu is too busy, my brain starts thinking too much, and I can't be mindful." They wanted big, clear buttons and a simple layout. No confusing menus.
• Make it Yours (The "Customization" Feature):
  People wanted to tweak the experience. They wanted to change the background colors, pick their own avatar (a digital character representing them), or choose their own music. One person said, "I want to customize my avatar so it looks like me, or maybe something totally different to give me a break from my illness."
• Safety First (The "Panic Button"):
  This was crucial. Mindfulness can sometimes bring up difficult emotions. The experts insisted on a "Safe Word" or "Safety Button." If a user feels overwhelmed, they need an instant way to stop the VR experience and get help, rather than being stuck in a virtual world while feeling bad.
• Nature vs. Noise:
  Most people loved natural sounds (waves, birds) and nature scenes. However, the experts warned that too much visual stimulation could be distracting. Sometimes, just listening to a calm voice with eyes closed is better than watching a fancy video.

The Result: A Prototype Blueprint

The study didn't build the final app yet, but it created a detailed blueprint (a software design document). It's like having the architectural plans for a house before you start building it.

The blueprint says:

• Design: Simple, clean, with natural colors.
• Features: Mood trackers, reminders, and a safety exit button.
• Content: Culturally appropriate voices, nature sounds, and the ability for users to create their own calming scenes using AI.

Why This Matters

Imagine if you could put on a headset, step into a peaceful, custom-made forest, and leave your diabetes worries at the door for 10 minutes. This study proves that people with diabetes want this kind of tool, and they know exactly what features would make it work for them.

The Takeaway:
Technology is powerful, but it only works if it fits the person using it. By mixing human creativity with Artificial Intelligence, and by listening to the people who actually need the help, the researchers have paved the way for a VR app that doesn't just look cool, but actually helps people breathe easier.

In short: They didn't just build a tool; they built a bridge between technology and human emotion, ensuring that when people with diabetes step into the virtual world, they step into a place of safety, not stress.

Technical Summary

1. Problem Statement

Diabetes Distress and Current Limitations:

• Context: Approximately 36% of adults with Type 2 Diabetes (T2D) experience "diabetes distress," a state of emotional overwhelm regarding daily self-care demands. This leads to poor glycaemic control, increased complications, and higher healthcare costs.
• Intervention Gap: While mindfulness-based interventions (MBIs) are effective, traditional face-to-face programs suffer from accessibility issues (geography, cost, scheduling) and lack personalization.
• Digital Limitations: Existing digital mindfulness apps often lack engagement, accountability, and personalized feedback, leading to low adherence.
• VR Potential & Gap: Virtual Reality (VR) offers immersive environments that can enhance focus and presence, potentially overcoming digital limitations. However, no existing VR mindfulness app is specifically tailored for T2D patients, and prior research lacks a true co-design approach where end-users are involved from the conceptualization phase. Most studies test pre-existing prototypes rather than building solutions based on user needs.

2. Methodology

The study employed a multi-method, two-phase feasibility design utilizing a novel framework called Artificial Intelligence-informed Experience-Based Co-Design (AI-EBCD). This framework adapts the standard Experience-Based Co-Design (EBCD) model by integrating artistic methods and Generative AI (GenAI) tools.

Phase 1: Expert Consultation & Market Review

• App Review: A review of five existing VR mindfulness apps (e.g., Headspace XR, TRIPP) using the Mobile App Rating Scale (MARS) to assess engagement, functionality, and aesthetics.
• Expert Interviews: Semi-structured interviews with 9 mindfulness practitioners (mean age 52, 78% experienced) to define key concepts, optimal timing/frequency, and virtual experience requirements.

Phase 2: Participatory Co-Design with End Users

• Participants: 13 adults with Type 2 Diabetes (mean age 32.6, 85% White British, 77% male) recruited via social media and diabetes charities.
• Process: Five sequential workshops were conducted:
  1. Workshop 1 (Focus Group): Discussed diabetes distress and mindfulness basics using emotional mapping.
  2. Workshop 2 (Evaluation): Participants evaluated existing VR apps via video/audio clips.
  3. Workshop 3 (Artistic Expression): Participants used drawing/sketching to visualize personalized mindfulness experiences.
  4. Workshop 4 (GenAI Integration): Participants used Generative AI tools (Picsart, MURF AI, Soundraw, Pictory) to create digital content (images, audio, video, music) for the app. Note: Due to copyright, these outputs informed design rather than being directly implemented.
  5. Workshop 5 (Prioritization): A Nominal Group Technique was used to rank and prioritize features identified in previous workshops.
• Data Analysis: Qualitative data (interviews, focus groups, artistic outputs) were analyzed using the Framework Approach (familiarization, coding, charting, mapping) in NVivo. Quantitative data were analyzed via descriptive statistics in Excel.

3. Key Contributions

• Methodological Innovation: Introduction of AI-EBCD, a novel participatory design framework that integrates Generative AI and artistic methods to allow non-technical users to rapidly prototype and visualize complex digital health concepts.
• Specific Design Specifications: The study produced a comprehensive Software Design Specification (SDS) for a VR mindfulness app tailored specifically for T2D patients, moving beyond generic mindfulness solutions.
• Safety & Ethics Framework: Identification of critical safety mechanisms (e.g., "safe words," support triggers) required for VR mindfulness to prevent distress in vulnerable populations.

4. Key Results & Findings

Two primary themes emerged regarding the desired VR application:

Theme 1: Design and Features

• Simplicity: Users demanded a simple, uncluttered interface with large fonts and clear navigation to avoid cognitive overload.
• Customization: High demand for personalization, including adjustable color palettes, avatars, notification settings, and the ability to tailor content (audio/visual) to individual preferences.
• Interactivity: Features such as mood trackers, gamification (quizzes/games), and social connectivity (sharing progress with "friends" or virtual forums) were requested to sustain engagement.
• Safety: A critical requirement for a "safety trigger" or button to exit the experience immediately if a user feels overwhelmed or distressed.

Theme 2: Content Strategy

• Audio: High preference for culturally appropriate, calm narration with neutral accents. While nature sounds were popular, some experts warned against excessive music that might distract from internal awareness.
• Visuals: Mixed views. Users liked nature-based imagery (landscapes, clouds) generated via GenAI. However, experts cautioned that overly pleasant visuals might invalidate a user's actual emotional state (e.g., pain or turmoil), suggesting a need for content that supports "being with" difficult emotions rather than just escaping them.
• Duration: Flexibility is key. Sessions should range from short (3-minute breathing spaces) for busy users to longer sessions (30+ mins) for those with time, accommodating varying life circumstances and neurodiversity.
• Content Updates: Regular updates and educational content regarding diabetes and mindfulness were requested to maintain long-term interest.

5. Significance and Future Directions

• Clinical Relevance: The study provides a blueprint for developing a VR intervention that addresses the specific emotional burdens of T2D, potentially improving self-management and reducing distress-related complications.
• Technological Advancement: Demonstrates the viability of using GenAI as a co-design tool, enabling patients to "speak" their design preferences through media creation rather than abstract description.
• Limitations: The study is a feasibility trial with a small, non-diverse sample (mostly White British, male, and relatively young). The participants were not experienced mindfulness practitioners, which may bias design toward novice needs.
• Next Steps: The authors call for larger, more diverse clinical trials to validate the efficacy of the prototype. Future work must rigorously test safety features to mitigate risks like cybersickness and psychological distress, and refine the balance between immersive visuals and the core principles of mindfulness (non-judgmental awareness).

Conclusion: This study successfully translated user needs into a technical specification for a VR mindfulness app, establishing that a co-designed, AI-informed approach is essential for creating effective, safe, and engaging digital health tools for chronic disease management.