Related Experiment Video
Updated: Mar 27, 2026
An Inertial Measurement Unit Based Method to Estimate Hip and Knee Joint Kinematics in Team Sport Athletes on the Field
Published on: May 26, 2020
Accuracy of Posture Allocation Algorithms for Thigh- and Waist-Worn Accelerometers
Charlotte L Edwardson1, Alex V Rowlands, Sarah Bunnewell
11Diabetes Research Centre, Leicester General Hospital, University of Leicester, Leicester, England, UNITED KINGDOM; 2NIHR Leicester-Loughborough Diet, Lifestyle and Physical Activity Biomedical Research Unit, Leicester, England, UNITED KINGDOM; 3Alliance for Research in Exercise, Nutrition and Activity, Sansom Institute for Health Research, Division of Health Sciences, University of South Australia, Adelaide, AUSTRALIA; 4National Centre for Sport and Exercise Medicine, School of Sport, Exercise and Health Sciences, Loughborough University, Loughborough, England, UNITED KINGDOM; 5School of Health Sciences, Stirling University, Stirling, Scotland, UNITED KINGDOM; 6University Hospitals of Leicester, Leicester General Hospital, Leicester, England, UNITED KINGDOM; and 7NIHR Collaboration for Leadership in Applied Health Research and Care, East Midlands, Leicester General Hospital, UNITED KINGDOM.
Wearable sensors like activPAL and ActiGraph accurately determine posture when worn on the thigh. However, waist-worn devices show poor accuracy for sitting detection, requiring caution when interpreting activity data.
Area of Science:
- Biomedical Engineering
- Physical Activity Measurement
- Wearable Technology
Background:
- Accurate measurement of physical activity and posture is crucial for health research.
- Wearable accelerometers are commonly used to assess posture, but their accuracy varies by device and placement.
- Proprietary and open-source algorithms exist for postural allocation, with differing performance characteristics.
Purpose of the Study:
- To compare the accuracy of proprietary and open-source postural allocation algorithms.
- To evaluate algorithms applied to activPAL, GENEActiv, and ActiGraph GT3X+ devices.
- To assess algorithm performance based on device placement (thigh vs. waist).
Main Methods:
- Thirty-four adults performed 16 activities (lying, sitting, upright) in a laboratory setting.
- Participants wore activPAL3, GENEActiv, and ActiGraph GT3X+ on the thigh, and ActiGraph on the hip.
- Direct observation was used to validate posture classification by the algorithms.
Main Results:
- Thigh-worn devices (activPAL, GENEActiv, ActiGraph) achieved high accuracy (≥91%) for all postures.
- The ActiGraph waist algorithm showed lower accuracy (58% for sitting).
- Specific sitting and lying postures with bent legs were misclassified by some thigh algorithms.
Conclusions:
- Postural allocation algorithms applied to thigh-worn devices demonstrate high accuracy.
- Waist-worn ActiGraph algorithm exhibits poor sitting detection accuracy.
- Caution is advised when inferring sitting time from waist-worn accelerometer data.
