Empowering Dyads of Older Adults With Mild Cognitive Impairment And Their Care Partners Using Conversational Agents

Title of the Paper

Empowering Dyads of Older Adults With Mild Cognitive Impairment And Their Care Partners Using Conversational Agents

Paper Information

• Subject Area: Human-Computer Interaction and Assistive Technology, exploring the application of conversational agents for older adults with cognitive impairments and their care partners
• Keywords: Conversational agents, voice assistants, older adults, mild cognitive impairment (MCI), care partners, smart environments, home health, assistive technology, user studies, psychological intervention

Research Background and Problem Statement

• Problems or Challenges:

  • Mild Cognitive Impairment (MCI) negatively impacts memory and executive functions in older adults, making it difficult for them to use complex technologies.
  • Older adults generally require more support and training with new technologies, and memory decline in MCI patients exacerbates this challenge.
  • Care partners of MCI patients often need to invest significant effort to support daily living, which can lead to caregiver burden.
  • How to leverage technology to enhance the independence of MCI patients while reducing the burden on care partners is a critical research question.

• Significance:

  • With global aging, the number of individuals with MCI is steadily increasing. Using technology to improve their quality of life has significant social implications.
  • Research in human-computer interaction suggests that voice assistant-based conversational agents (e.g., Google Home) can provide cognitive support through "hands-free" and "screen-free" interaction, potentially replacing traditional touchscreen or display-based designs.

• Related Work:

  • Existing studies have demonstrated that conversational agents can assist users with home automation, scheduling reminders, and more, but there is limited research on the long-term usage patterns of MCI patients and their care partners.
  • Previous research has primarily focused on basic functionality and technical implementation, with less attention to psychological and functional support for special populations such as those with cognitive impairments.

Proposed Solution

• Proposed Methods and Solutions:

  • This study designed and conducted a 10-week field study to observe how 10 dyads of MCI patients and their care partners used the Google Home Hub (GHH) conversational agent in their homes.
  • Participants were guided to explore various functions of the home conversational agent and actively design and use features such as scheduling and reminder tools to enhance independent interaction.

• Innovations:

  • Proposed a three-tier interaction framework that categorizes conversational agent interaction types into "basic functions," "personalized functions," and "skill extensions," and analyzed difficulty levels based on users' cognitive abilities.
  • Emphasized the role of care partners in providing indirect support to MCI patients through "interaction scaffolding" (e.g., setting reminders, customizing parameters) in daily life.
  • Introduced "empowerment" as a goal, analyzing psychological support and technological independence for this population.

• Implementation Steps and Techniques:

  • Deployed Google Home Hub devices and completed pairing with users' phones and Google accounts, enabling voice recognition and personalized scheduling.
  • Provided a series of phased training materials, including "quick command guides," instructional videos, and extended feature lists, and added functionality descriptions based on user feedback.
  • Collected all user interaction data (3878 logs) for quantitative analysis and conducted qualitative experience data collection through Zoom interviews and online surveys.

Research Findings

• Specific Findings:

  • The results summarized interaction patterns and types from 3878 user behaviors of MCI patients and their care partners, showing that care partners used the system more frequently and engaged more with personalized functions.
  • User interactions were primarily concentrated on basic functions (79%), followed by personalized functions (approximately 20%) and skill extensions (about 2%).
  • Qualitative interviews revealed that some care partners reduced their caregiving burden by using features like reminders and scheduling tools.
  • Users provided positive feedback on how conversational agents improved daily independence and alleviated loneliness, while a small portion expressed negative emotions about the system's complexity.

• Comparisons and Advantages:

  • Compared to previous studies, this research more comprehensively explored the application of conversational agents in a specific population (MCI patients and care partners), particularly in psychological empowerment and collaborative task management.
  • This study was the first to detail the impact of cognitive load at different interaction levels on the usage rate of conversational agent functions.

• Experimental and Evaluation Results:

  • Observations showed that care partners' "interaction scaffolding" behaviors accounted for 29% of their total interactions, significantly reducing the burden on patients for queries and memory-related issues.
  • User surveys indicated that most participants hoped for further optimization of conversational agents in areas such as reminders and danger alerts.

• Limitations and Future Directions:

  • Limitations:
    • The study focused on exploring the functionality of conversational agents without deeply investigating the specific long-term effects on cognitive functions.
    • Some users found complex features (e.g., skill extensions) difficult to master, suggesting the need for more simplified interaction designs.
    • Many activities moved online due to the COVID-19 pandemic, which may have caused overuse or interference with the results.
  • Future Directions:
    • Explore richer sensor integrations (e.g., danger monitoring).
    • Develop automated "wake word" optimization to reduce reliance on specific voice commands.
    • Provide more systematic training models with personalized learning paths.