Personalized Multilayer Daily Life Profiling Through Context Enabled Activity Classification and Motion...

Area of Science:

• Rehabilitation Engineering
• Wearable Technology
• Activity Recognition

Background:

• Monitoring patient activity is crucial for assessing rehabilitation progress and adherence to exercise recommendations.
• Current methods often lack detailed, real-world activity data for physically disabled individuals.

Purpose of the Study:

• To develop and validate an end-to-end system for comprehensive daily activity profiling in physically disabled individuals.
• To enable healthcare professionals to monitor patient compliance and performance in community settings.

Main Methods:

• Utilized a physical activity prescription model with MEMS inertial sensors and mobile devices.
• Implemented a context-guided activity monitoring approach with categorical location context.
• Integrated motion reconstruction, visualization, and metrics computation for layered reporting.

Main Results:

• The system provides multi-level granularity: location categories, activities within locations, and detailed motion analysis.
• A novel location context detection algorithm using WiFi-augmented GPS enhanced accuracy.
• Experimental field evaluations confirmed the system's accuracy.

Conclusions:

• The developed system offers a robust solution for in-community activity profiling for physically disabled individuals.
• This technology facilitates personalized feedback and improved healthcare monitoring, supporting rehabilitation goals.