Related Experiment Video
Updated: May 20, 2026
11:21
Methodology for Establishing a Community-Wide Life Laboratory for Capturing Unobtrusive and Continuous Remote Activity and Health Data
Published on: July 27, 2018
Enabling large-scale ground-truth acquisition and system evaluation in wireless health
James Y Xu1, Greg J Pottie, William J Kaiser
1Department of Electrical Engineering, University of California, Los Angeles, Los Angeles, CA 90024, USA. jyxu@ucla.edu
IEEE Transactions on Bio-Medical Engineering
|July 18, 2012
Summary
This study addresses challenges in large-scale activity monitoring for fitness and healthcare. It introduces a voice-powered system and a comprehensive database to enable robust data collection and system comparison.
Area of Science:
- Human-computer interaction
- Biomedical informatics
- Wearable technology
Background:
- Large-scale activity monitoring is crucial for fitness, healthcare management, and diagnostics.
- Existing research primarily focuses on motion classification accuracy, neglecting scalability challenges in community settings.
Purpose of the Study:
- To introduce and address the problem of scaling activity monitoring systems for large communities.
- To develop methods for robust large-scale ground-truth acquisition.
- To establish a common database for comparing different activity monitoring systems.
Main Methods:
- A voice-powered mobile acquisition system was developed for efficient data collection.
- Efficient annotation tools were integrated to streamline the ground-truth acquisition process.
- An extendable, online, searchable activity database was created.
Main Results:
- The database contains 331 datasets, totaling over 700 hours of activity data.
- Data was collected using 8 different sensing modalities.
- 15 distinct human activities were recorded and annotated.
Conclusions:
- The developed system and database facilitate robust large-scale ground-truth acquisition.
- This work provides a foundation for standardized comparison of activity monitoring systems.
- Addressing scalability is essential for advancing the application of activity monitoring in real-world settings.
