RemiHaven: Integrating "In-Town" and "Out-of-Town" Peers to Provide Personalized Reminiscence Support for Older Drifters

Research Background and Issues

• What problems or challenges did the authors identify?

  • With increasing social mobility and the growing prominence of aging issues, more elderly individuals in China are relocating to new cities to live with their children. These elderly migrants, referred to as the "Elderly Drifters," often face psychological issues such as loneliness, loss, and depression due to a lack of social connections and difficulties in cultural adaptation.
  • Traditional nostalgia intervention methods have proven effective in improving the mental health and cognitive functions of the elderly. However, these methods typically rely on tangible nostalgic materials (e.g., old photographs, memorabilia), which are often inaccessible to the Elderly Drifters due to relocation. Additionally, this group lacks sufficient social resources to support traditional nostalgia interventions.

• Why is this issue important?

  • The psychological issues faced by the Elderly Drifters not only affect their personal well-being but may also impose additional burdens on their families and society.
  • This study focuses on how technology can address the psychological and social support gaps caused by relocation, while also exploring new design approaches to provide mental health support for the rapidly growing population of Elderly Drifters.

• Research Motivation and Related Work

  • In the field of human-computer interaction, various tools (e.g., VR, touchscreens) have been explored to support nostalgia interventions. However, these methods primarily target elderly individuals with cognitive impairments, and their applicability to the Elderly Drifters remains to be validated.
  • The rise of multimodal large language models (MLLMs) offers new opportunities to address the lack of nostalgic materials and personalized needs of the Elderly Drifters by generating interactive content and providing personalized experiences.

Solution

• What methods or solutions did the authors propose?

  • The authors designed a nostalgia support tool called "RemiHaven," which incorporates "In-Town" and "Out-of-Town" character agents to provide personalized nostalgic experiences for the Elderly Drifters.
  • The tool leverages MLLM technology to generate personalized multimodal content (e.g., conversations, images, and text) that evokes users' memories and creates a "Life Storybook" for preservation and sharing.

• What are the innovative aspects of this solution?

  • The introduction of two virtual characters—"In-Town Peers" and "Out-of-Town Peers"—to cater to different nostalgic needs is a novel design specifically tailored to the nostalgia characteristics of the Elderly Drifters.
  • The system generates personalized multimodal content related to users' memories by integrating their background information and conversational context.
  • It provides iterative interaction and user modification capabilities for the generated content, making the memory recall process more flexible and engaging.

• What are the implementation steps and key technologies used?

  • User Configuration Module: Collects users' basic background information (e.g., gender, age, hometown) and integrates relevant geographical knowledge for personalized content generation.
  • Two Interaction Modes:
    • "In-Town Peers": Simulates acquaintances from the user's hometown to actively guide memory recall.
    • "Out-of-Town Peers": Acts as an outsider who listens to the user's memories, offering space for free expression.
  • Generation and Modification Features: Utilizes multimodal generation technologies (e.g., DALL·E and ChatGPT-4) to allow users to create and adjust nostalgic content (images, text), culminating in an exportable PDF Life Storybook.

Research Outcomes

• What specific outcomes were achieved?

  • Evaluations showed that RemiHaven significantly enhanced the nostalgic experiences of the Elderly Drifters, helping users recall past events and positively impacting their emotions.
  • The multimodal generated content (images and text) and the nostalgic character mode design were widely appreciated by users.
  • The "Life Storybook" generation and export functionality were deemed highly beneficial for memory preservation and sharing.

• What advantages does it have compared to existing solutions?

  • It provides personalized nostalgia support tailored to the unique needs of the Elderly Drifters, whereas existing tools often overlook the challenges of material loss and cultural disconnection caused by relocation.
  • The system supports flexible control of nostalgic interaction modes (a combination of guided and free-form), accommodating different user preferences.

• What were the experimental or evaluation results?

  • Among the 10 Elderly Drifter participants tested, the "In-Town Peers" interaction mode was widely recognized, while the "Out-of-Town Peers" mode performed less effectively when users needed to switch topics quickly.
  • The voice interaction feature was considered convenient by most users, though it struggled with recognizing speech from participants with unclear pronunciation.
  • The image and text generation features performed well overall but require further optimization to better understand users' hometown cultural contexts.

• Limitations and Future Directions

  • The user base was relatively limited, with tests conducted only on Elderly Drifters from a single city. Future research should expand to include more diverse cultural and regional backgrounds.
  • The system currently relies on manually optimized prompt engineering and has not yet achieved adaptive dynamic interaction.
  • The long-term effects of the system have not been fully validated. Future studies should test its use in real-life scenarios to assess whether it provides sustained psychological and social benefits.
  • It is recommended to explore additional interaction methods (e.g., image input) and integrate offline activities to enhance social connections.