Related Experiment Video
Updated: May 22, 2025
Measuring the Functional Abilities of Children Aged 3-6 Years Old with Observational Methods and Computer Tools
Published on: June 20, 2020
Monitoring of child-specific activities in ambulatory children with and without developmental disabilities
Barbara Engels1,2,3, Manon A T Bloemen4,5,6, Richard Felius1,7
1Research Centre for Healthy and Sustainable Living, Research Group Lifestyle and Health, Utrecht University of Applied Sciences, Utrecht, The Netherlands.
Insights
This study validated an ankle-worn activity monitor (AM-p) for children. The prototype accurately distinguished between stationary, cycling, and locomotion activities in real-world settings, aiding pediatric physical activity assessment.
Area of Science:
- Pediatric health
- Biomedical engineering
- Kinesiology
Background:
- Children's physical activity is crucial for health, yet objective measurement in natural environments is challenging.
- Activity monitors (AM) offer potential for assessing physical activity, but validation in real-world settings is limited.
- Existing AM validation studies often lack ecological validity, necessitating research in children's everyday environments.
Purpose of the Study:
- To evaluate the criterion validity of a prototype ankle-worn activity monitor (AM-p).
- To assess the AM-p's ability to differentiate physical activities in a natural setting for children aged 2-19 years.
- To determine the accuracy of the AM-p for children with and without developmental disabilities.
Main Methods:
- A cross-sectional community-based study involving ambulatory children (2-19 years) with and without developmental disabilities.
- Children wore the AM-p on the ankle while performing activities filmed using a gold-standard approach.
- Machine learning techniques were employed to analyze AM-p data against synchronized video-labeled activity epochs (stationary, cycling, locomotion).
Main Results:
- The AM-p model achieved 82% accuracy, 78% recall, 75% precision, and 75% F1 score in differentiating three activity categories.
- Analysis included 93 children (mean age 11 years, 55% girls), with 28 having developmental disabilities.
- Older children (>13 years) with typical development showed higher assessment accuracy compared to younger children and those with developmental disabilities.
Conclusions:
- The ankle-worn AM-p demonstrates good accuracy (82%) in distinguishing between stationary, cycling, and locomotion activities in children.
- The AM-p's ability to assess a heterogeneous group of ambulatory children, including those with developmental disabilities, supports its clinical utility.
- This validated AM-p holds promise for enhancing physical activity assessment by pediatric healthcare professionals in clinical settings.
Background:
Pediatric healthcare professionals facilitate children to enhance and maintain a physically active lifestyle. Activity monitors (AM) can help pediatric healthcare professionals assess physical activity in everyday life. However, validation research of activity monitors has often been conducted in laboratories and insight into physical activity of children in their own everyday environment is lacking. Our goal was to study the criterion validity of a prototype AM (AM-p) model in a natural setting.
Methods:
Cross-sectional community-based study with ambulatory children (2-19 years) with and without developmental disability. Children wore the AM-p on the ankle and were filmed (gold standard) while performing an activity protocol in a natural setting. We labelled all videos per 5-second epoch with individual activity labels. Raw AM-p data were synchronized with activity labels. Using machine learning techniques, activity labels were subdivided in three pre-defined categories. Accuracy, recall, precision, and F1 score were calculated per category.
Results:
We analyzed data of 93 children, of which 28 had a developmental disability. Mean age was 11 years (SD 4.5) with 55% girls. The AM-p model differentiated between 'stationary', 'cycling' and 'locomotion' activities with an accuracy of 82%, recall of 78%, precision of 75%, and F1 score of 75%, respectively. Children older than 13 years with typical development can be assessed more accurately than younger children (2-12 years) with and without developmental disabilities.
Conclusion:
The single ankle-worn AM-p model can differentiate between three activity categories in children with and without developmental disabilities with good accuracy (82%). Because the AM-p can be used for a heterogenous group of ambulatory children with and without developmental disabilities, it may support the clinical assessment for pediatric healthcare professionals in the future.
