Related Experiment Video
Updated: Mar 18, 2026
06:51
Physical Activity Measurement in Children Accepting Table Tennis Training
Published on: July 27, 2022
2.6K
Predictive Validity of a Thigh-Worn Accelerometer METs Algorithm in 5- to 12-Year-old Children
Christiana M van Loo1, Anthony D Okely, Marijka Batterham
1Early Start Research Institute and the Illawarra Health and Medical Research Institute, Faculty of Social Sciences, Wollongong, NSW, Australia.
Journal of Physical Activity & Health
|July 9, 2016
Summary
The activPAL3 algorithm accurately classified moderate-to-vigorous physical activity (MVPA) in children but overestimated sedentary behavior (SB) and light physical activity (LPA). Validation showed good MVPA classification despite wide individual errors.
Area of Science:
- Pediatric physical activity monitoring
- Algorithm validation in children
- Metabolic equivalent estimation
Background:
- Accurate measurement of physical activity in children is crucial for public health.
- Wearable accelerometers are commonly used but require validation.
- The activPAL3 device's algorithm for predicting metabolic equivalents (TAMETs) and classifying physical activity intensity needs evaluation in pediatric populations.
Purpose of the Study:
- To validate the activPAL3 algorithm for predicting metabolic equivalents (TAMETs) and classifying moderate-to-vigorous physical activities (MVPA) in children aged 5–12 years.
- To compare TAMETs with criterion measures of energy expenditure and physical activity intensity.
- To assess the accuracy of the activPAL3 algorithm across different activity types.
Main Methods:
- Fifty-seven children (9.2 ± 2.3 years) performed 14 activities, including sedentary behaviors (SB), light physical activity (LPA), and MVPA.
- Indirect calorimetry (IC) served as the criterion measure for energy expenditure.
- Analyses included equivalence testing, Bland-Altman procedures, and receiver operating curve analysis (ROC-AUC).
Main Results:
- Group-level TAMETs were equivalent to IC for specific activities like e-gaming, writing/coloring, and standing (P < .05).
- TAMETs generally overestimated SB (7.9 ± 6.7%) and LPA (1.9 ± 20.2%), and underestimated MVPA (27.7 ± 26.6%).
- MVPA classification accuracy was good (ROC-AUC = 0.86), but limits of agreement were wide, indicating significant individual variability.
Conclusions:
- The activPAL3 algorithm demonstrated acceptable accuracy for classifying MVPA in children.
- The algorithm showed overestimation for SB and LPA and underestimation for MVPA at the group level.
- Wide individual error limits suggest caution when interpreting absolute energy expenditure values for individual children.
