Cross-validation of waist-worn GENEA accelerometer cut-points
Whitney A Welch1, David R Bassett, Patty S Freedson
11Department of Kinesiology, Recreation, and Sport Studies, University of Tennessee, Knoxville, TN; 2Department of Kinesiology, University of Massachusetts, Amherst, MA; and 3Department of Health Sciences, Northeastern University, Boston, MA.
Medicine and Science in Sports and Exercise
|February 6, 2014
Summary
The waist gravity estimator of normal everyday activity (GENEA) showed low accuracy (55.3%) in classifying lifestyle activity intensity. Further validation is needed for reliable physical activity intensity assessment.
Area of Science:
- Physical activity monitoring
- Biomedical engineering
- Exercise physiology
Background:
- Wearable sensors are increasingly used to monitor physical activity.
- Accurate assessment of activity intensity is crucial for public health and clinical applications.
- Existing methods require validation across diverse lifestyle activities.
Purpose of the Study:
- To evaluate the classification accuracy of the GENEA's waist-worn cut-points.
- To assess the prediction of intensity categories during various lifestyle activities.
- To cross-validate the GENEA's performance against direct measures of oxygen uptake.
Main Methods:
- Participants engaged in 7 lifestyle activities (home/office, ambulatory, sport).
- The GENEA was worn at the waist, with continuous oxygen uptake measured by Oxycon mobile.
- Statistical analyses included chi-squared tests, cross-tabulations, and sensitivity/specificity analyses.
Main Results:
- Spearman correlation between GENEA signal and MET values was 0.73.
- Overall classification accuracy for the GENEA was 55.3%, improving to 58.3% excluding stationary cycling.
- Sensitivity ranged from 0.244 to 0.958, and specificity from 0.576 to 0.943 across intensity categories.
Conclusions:
- The GENEA cut-points demonstrated low overall accuracy (55.3%) for classifying intensity across 14 lifestyle activities.
- This cross-validation highlights limitations in the current GENEA cut-points for real-world activity monitoring.
- Further refinement of algorithms and cut-points is recommended for improved accuracy.
