PD-Insighter: A Visual Analytics System to Monitor Daily Actions for Parkinson's Disease Treatment

Title of the Paper

PD-Insighter: A Visual Analytics System to Monitor Daily Actions for Parkinson’s Disease Treatment

Paper Information

• Field of Study: Application of data visualization and immersive analytics in monitoring daily life activities for Parkinson’s disease
• Keywords: Data visualization, Parkinson’s disease, immersive analytics, motion analysis, medical assistance, hybrid interface, motion data analysis

Research Background and Issues

• Identified Problems or Challenges:

  • Daily motor performance of Parkinson’s disease patients can only be assessed through brief clinical observations or subjective self-reports, which are prone to memory bias and lack sufficient information.
  • There is a lack of appropriate tools for analyzing patients’ movement patterns in daily environments to identify critical motor deficiencies during treatment.
  • Current wearable sensor data provides isolated information, making it difficult to comprehensively analyze patient environment, actions, and body motion data.

• Importance:

  • Understanding body movement patterns and environmental factors in activities of daily living (ADLs) is crucial for the treatment of Parkinson’s disease patients.
  • The effectiveness of physical therapy depends on tailored treatment plans, which require a comprehensive understanding of patients’ long-term motor performance.

• Research Motivation and Related Work:

  • Existing studies focus on clinical record-based motion pattern analysis, lacking exploration of long-term, multi-scale motion data in home environments.
  • Current methods often emphasize single dimensions, such as action types or specific time points, neglecting the importance of integrated analysis of actions, dynamics, and environment.
  • Immersive analytics has demonstrated potential in enhancing navigation and understanding of complex spatial data.

Solution

• Proposed Method or Solution:

  • Developed PD-Insighter, a hybrid analytics system combining desktop and augmented reality (AR) platforms, which analyzes patients’ daily motion data through multidimensional visualization tools to quickly identify critical motor deficiencies.
  • The system includes an “overview dashboard” for high-level trend analysis of actions and variables, and an “immersive replay” feature for in-depth exploration of body-environment interactions at specific time points.

• Innovations:

  • Hybrid Platform Design: Combines the efficiency of desktop data analysis with the spatial realism of AR for fine-grained analysis.
  • Data Integration and Visualization: Provides comprehensive data representation through body variables, action labels, and 3D environmental reconstruction.
  • Task-Driven Design: System functionalities and visualization designs are based on core clinical tasks summarized from in-depth interviews with therapists.

• Implementation Steps and Key Technologies:

  • Data Processing:
    • Extract key body variables (e.g., trunk angle, limb positions, weight distribution) from wearable inertial measurement units (IMUs) and video camera data.
    • Use computer vision methods to identify action events in daily life (e.g., walking, sitting, standing).
    • Reconstruct environmental videos into 3D environment grids to enhance contextual presentation.
  • Interface Design:
    • Present motion and activity data in the overview dashboard using heatmaps, bar charts, etc., with variable and event filtering capabilities.
    • Recreate patient body skeleton and surrounding environment in AR, enabling therapists to freely analyze motion details based on depth and perspective.

Research Outcomes

• Specific Results:

  • Therapists can quickly identify abnormal motion patterns and analyze their environmental and variable contexts through the overview dashboard.
  • The immersive replay feature provides realistic spatial perception, allowing therapists to observe patients’ body movements from multiple angles and infer potential triggering factors.
  • User studies with six rehabilitation experts demonstrated the system’s capability to support rapid and effective insights into patient motion data.

• Comparison with Existing Solutions and Advantages:

  • Offers multi-scale analysis capabilities (from global motion trends to specific action details), surpassing traditional visualization methods that only focus on fragmented actions.
  • AR technology enables users to perceive 3D motion contexts more intuitively compared to traditional 2D video playback.
  • Dynamic aggregation and anomaly highlighting techniques significantly improve the efficiency of detecting abnormal motion patterns in long-term data.

• Experimental or Evaluation Results:

  • Participating rehabilitation experts successfully completed the following tasks using the system:
    • Understanding patients’ overall motion trends (e.g., “excessive sedentary time”).
    • Quickly identifying critical motor deficiency events, such as gait freezing, imbalanced arm swinging, and difficulty in sitting or standing.
    • Using the replay feature to verify motor deficiencies while understanding environmental triggering factors.
  • User feedback indicated significant potential for clinical analysis, especially in time-constrained diagnostic settings.

• Limitations and Future Directions:

  • Limitations:
    • The current system uses static action labels and variables, lacking fully automated action detection mechanisms.
    • Patient data privacy concerns have not been fully addressed.
    • Long-term, sparser home-collected data has not been tested; future methods need to expand to handle large-scale datasets.
  • Future Directions:
    • Automate action labeling through deep learning to further reduce manual intervention.
    • Develop privacy protection technologies (e.g., facial blurring, environment reconstruction reduction) to meet real-world deployment needs.
    • Explore extensions to other fields of motion data analysis, such as stroke rehabilitation or sports performance analysis.