Exploring the Accessible Design of Virtual Reality Games of Different Intervening Purposes with People with Dementia at Different Stages

Technical Summary: Exploring the Accessible Design of Virtual Reality Games of Different Intervening Purposes with People with Dementia at Different Stages

Problem Statement
Virtual Reality (VR) is an emerging non-pharmaceutical intervention for dementia, offering safe, motivating, and natural interaction contexts. However, existing literature largely treats people with dementia as a homogeneous group or focuses on a single disease stage (e.g., Mild Cognitive Impairment). Consequently, there is insufficient discussion on the accessible design of VR games when applied to users across the spectrum of cognitive decline: Subjective Cognitive Decline (SCD), Mild Cognitive Impairment (MCI), and Alzheimer's Dementia (AD). Furthermore, current research often fails to differentiate between VR games designed for specific intervening purposes: cognitive training (serious games) versus recreational exercise (entertainment). The study addresses the need to understand the feasibility, limitations, and accessibility barriers of these distinct game types for users at different stages of cognitive impairment, moving beyond simple device wearability to consider memory load, navigation burden, emotional safety, and task familiarity.

Methodology
The authors conducted a qualitative user study involving 11 participants recruited from the Chinese Imaging, Biomarkers and Lifestyle (CIBL) Study of Ageing Project. The cohort included:

• 4 participants with Alzheimer's Dementia (AD)
• 4 participants with Mild Cognitive Impairment (MCI)
• 3 participants with Subjective Cognitive Decline (SCD)

Participants engaged with two distinct VR applications on an Oculus Quest 2:

1. A Custom Cognitive Training Game ("Grocery Shopping"): Developed via participatory design with clinical professionals and people with dementia. It targets memory, spatial navigation, and calculation. The game features three difficulty levels tailored to AD, MCI, and SCD, utilizing steering locomotion and varying interaction complexity (e.g., number of buttons, memory load).
2. A Commercial Recreational Game ("First Steps" Dancing): A pre-existing application where users mimic a robot's gestures, focusing on physical exercise and social interaction.

Procedure:

• Participants played both games for 15 and 5 minutes respectively (order randomized) while seated in a swivel chair for safety.
• Data collection included video recordings of physical and virtual interactions, semi-structured interviews based on the Technology Acceptance Model 3 (TAM3) (assessing Perceived Usefulness, Ease of Use, Enjoyment, and Behavioral Intention), and caregiver feedback.
• Analysis: Data underwent thematic analysis using NVivo 12. The process involved inductive coding, creating open codes, and refining them into three primary thematic maps based on observation notes, participant feedback, and caregiver input.

Key Results and Themes
The thematic analysis yielded three core findings regarding the accessibility and utility of VR for dementia:

1. Limitations of Single-Purpose Interventions:

   • Games with a single intervening purpose (either purely cognitive training or purely recreational) have differentiated usefulness depending on the user's stage.
   • Users with AD often found the cognitive training game too burdensome or boring, preferring the low-learning-cost recreational game for physical stimulation. Conversely, users with SCD/MCI valued the cognitive training for its mental exercise but sometimes found the recreational game too simple or lacking in social depth.
   • Suitability is influenced by daily life interests, subjective perceived effectiveness, and current cognitive capacity.

2. Crucial Need for Personalized Difficulty Matching:

   • Static difficulty levels are insufficient. For AD users, even "Level 1" of the training game could be overwhelming if not perfectly matched, leading to abandonment.
   • For SCD/MCI users, a game that is too simple fails to provide a sense of effectiveness or challenge.
   • The study highlights that individual differences exist even within the same diagnostic stage, necessitating dynamic, personalized difficulty strategies rather than fixed tiers.

3. Social Interaction in Recreational VR Stimulates Positive Emotions:

   • The dancing game, featuring a virtual robot, successfully triggered positive emotional responses (joy, companionship) across all stages.
   • For AD users, the robot acted as a social object that reduced isolation and prompted physical movement (e.g., mimicking gestures) even in users with a decade-long history of apathy.
   • Users expressed a desire for more interactive capabilities (e.g., talking to the robot), suggesting that social integration is a key driver of engagement.

Significance and Design Implications
The paper posits that accessible VR design for dementia must evolve from fixed intervention models to flexible, multi-dimensional systems. The authors extract three specific design implications for future work:

1. Multi-Component/Multi-Dimensional Strategies: Future systems should combine cognitive training and recreational elements. This allows for a "work and rest" cycle where recreational games motivate users to engage in cognitive training, addressing both physical and mental outcomes. Caregivers should be empowered as active mediators to switch between modes based on user fatigue and interest.
2. AI-Empowered Difficulty Matching: To address the limitations of static difficulty, the authors propose using Artificial Intelligence to adapt game difficulty in real-time. This could utilize metrics such as speed thresholds, success rates, reaction times, and error types to adjust challenges, thereby reducing failure experiences and protecting user confidence.
3. Emotionally Intelligent Virtual Agents: Recreational VR should integrate agents with personality-driven behaviors and emotional intelligence. These agents should provide companionship, gesture demonstration, and affective encouragement to enhance social interaction and reduce the pressure of the VR environment.

Limitations and Future Work
The authors acknowledge this is a preliminary qualitative study with a small sample size, partly due to pandemic restrictions and consent limitations. The comparison between a custom prototype and a commercial game limits controlled clinical effectiveness claims. Future work aims to expand the sample size for quantitative conclusions, develop adaptive VR prototypes to verify the AI-driven difficulty matching, and conduct longer-term studies with richer interaction logs.