“I Don’t Want to Become a Number’’: Examining Different Stakeholder Perspectives on a Video-Based Monitoring System for Senior Care with Inherent Privacy Protection (by Design).

Literature Title

"I Don’t Want to Become a Number": Examining Different Stakeholder Perspectives on a Video-Based Monitoring System for Senior Care with Inherent Privacy Protection (by Design)

Literature Information

• Subject Area: Intelligent Monitoring Technology, Privacy Protection, Senior Care
• Keywords: Active and Assisted Living, Video Monitoring System, Privacy Concerns, Older Adults, Ageing, Qualitative Study

Research Background and Issues

• Identified Problems or Challenges:

  1. With the acceleration of global aging, how to leverage technology to support the safe and independent living of older adults has become a critical issue.
  2. While video monitoring technology is powerful in health monitoring, it poses privacy leakage concerns.
  3. Empirical studies on user experience and privacy concerns regarding video monitoring systems are relatively scarce.
  4. There is a lack of in-depth research on balancing privacy protection and user acceptance in real-world deployment scenarios of video monitoring systems.

• Significance:

  1. Improving the quality of senior care.
  2. Alleviating labor shortages and supporting home-based living and operations in medical institutions for older adults.
  3. Exploring the optimal balance between privacy and safety to promote technology acceptance and implementation.

• Research Motivation and Related Work:

  1. This study is based on the "Privacy by Design" concept, conducting an extensive review of visual privacy protection mechanisms.
  2. It introduces the "privacy-by-context" approach, dynamically adjusting privacy levels based on the environment and user preferences.
  3. While current efforts address balancing privacy protection and medical safety needs in video monitoring, comprehensive studies incorporating actual user needs are lacking.

Solution

• Proposed Method:

  1. Develop and evaluate a video monitoring system with privacy protection features.
  2. The system includes multiple privacy filtering modes (blurring, pixelation, avatar substitution, complete obfuscation) and "contextual privacy" features.
  3. The system can dynamically adjust protection levels, such as displaying real images to caregivers in the event of a fall.

• Innovations:

  1. Introduces the "privacy-by-context" approach, dynamically adjusting privacy protection based on users, environments, and activity contexts.
  2. Utilizes lightweight deep learning models, independent of specific hardware, enhancing real-time performance and scalability.
  3. Extends services to intuitive privacy adaptation functions in video monitoring, providing older users with more precise and user-friendly solutions.

• Implementation Steps and Key Technologies:

  1. Use the "Fast and Light DensePose" deep learning algorithm to analyze human posture and actions in videos.
  2. Develop four privacy protection filters: blurring, pixelation, avatar substitution, and obfuscation.
  3. Automatically trigger privacy cover removal mechanisms in emergencies, such as fall detection, to display real scenes for emergency care.
  4. Deploy the system on embedded devices (NVIDIA Jetson Xavier NX) combined with high-resolution cameras, supporting online computation.

Research Outcomes

• Specific Outcomes:

  1. The video monitoring system met the fundamental privacy and safety needs of different stakeholders, particularly the reliability of privacy protection and the intuitive safety features for older users.
  2. The "Avatar" filter was the most popular among participants due to its anonymity and strong privacy protection; the fall event handling feature was widely recognized.
  3. A comprehensive analysis was conducted on the preferences and acceptance conditions of different user groups (older adults, caregivers, and technical staff) regarding the filters.

• Advantages Compared to Existing Solutions:

  1. Improved visual privacy protection methods allow users to achieve a better balance between privacy and safety.
  2. Simplified traditional video monitoring solutions reliant on multiple hardware components, introducing flexible designs with personalized features.

• Experimental or Evaluation Results:

  1. A total of 29 participants (13 older users, 16 caregivers and technical staff) provided detailed feedback during user testing.
  2. The system received positive evaluations in terms of simplified human-computer interaction, visual effects, and privacy mode design.
  3. Despite concerns about costs and practical implementation issues (e.g., unstable network connections), users showed high acceptance of the technology.

• Limitations and Future Directions:

  1. The study was limited to short-term usage scenarios, necessitating further research on long-term deployment to assess real-world effectiveness.
  2. Explore user preferences and the dynamic balance between privacy and safety in private areas (e.g., bedrooms, bathrooms).
  3. Promote co-design methods based on deep user involvement and learning, including broader user groups and practical research design methods (e.g., participatory design, living labs).