Accelerometer-Based Assessment of Physical Activity in Young Children

Accelerometers have become the cornerstone of objective measurement of physical activity in young children, offering high-resolution data on movement patterns across daily living. Worn at the waist, ankle or wrist, these devices capture acceleration with fine temporal granularity, enabling estimation of energy expenditure and classification of behaviours into sedentary, light and moderate-to-vigorous intensity categories. Processing of raw acceleration signals relies on epoch segmentation and application of intensity thresholds, or cut points, derived from calibration studies. However, standard cut-point approaches can lack generalisability across developmental stages and diverse populations. Recent advances include machine-learning frameworks that identify activity states directly from raw data, and bespoke threshold derivation for pre-ambulatory infants. Methodological consensus on device placement, epoch length and non-wear criteria remains elusive, complicating cross-study comparison. Despite these challenges, accelerometry underpins large-scale surveillance of adherence to international guidelines in preschoolers, informs early-life interventions and supports policy aimed at reducing sedentary time and promoting age-appropriate movement behaviours worldwide.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Accelerometer-Based Assessment of Physical Activity in Young Children publication trend

The graph below shows the total number of articles in accelerometer-based assessment of physical activity in young children across all publications each year (not limited to Nature Index journals).

Technical terms

Accelerometer: A wearable sensor that records acceleration forces to quantify movement intensity and frequency.

Cut point: A threshold value applied to processed accelerometer data to classify epochs into intensity categories such as sedentary or moderate-to-vigorous.

Epoch: A defined time interval, typically ranging from two seconds to one minute, over which accelerometer output is summarised.

Moderate-to-vigorous physical activity (MVPA): Movement behaviour characterised by acceleration above a predefined threshold, corresponding to increased energy expenditure.

Unsupervised machine learning: A computational approach that identifies patterns or states in unlabelled data without relying on predetermined categories.

Hidden semi-Markov model: A statistical method that segments sequential data into discrete states based on probabilistic transition and duration characteristics.

References

1. Using unsupervised machine learning to quantify physical activity from accelerometry in a diverse and rapidly changing population. PLOS Digital Health (2023).
2. Adherence to the World Health Organization’s physical activity recommendation in preschool-aged children: a systematic review and meta-analysis of accelerometer studies. International Journal of Behavioral Nutrition and Physical Activity (2023).
3. Accelerometer Thresholds for Estimating Physical Activity Intensity Levels in Infants: A Preliminary Study. Sensors (2024).