Abstract:Accelerometers are widely used to measure physical activity behaviour, including in children. The traditional method for processing acceleration data uses cut points to define physical activity intensity, relying on calibration studies that relate the magnitude of acceleration to energy expenditure. However, these relationships do not generalise across diverse populations and hence they must be parametrised for each subpopulation (e.g., age groups) which is costly and makes studies across diverse populations and over time difficult. A data-driven approach that allows physical activity intensity states to emerge from the data, without relying on parameters derived from external populations, offers a new perspective on this problem and potentially improved results. We applied an unsupervised machine learning approach, namely a hidden semi-Markov model, to segment and cluster the raw accelerometer data recorded (using a waist-worn ActiGraph GT3X+) from 279 children (9–38 months old) with a diverse range of developmental abilities (measured using the Paediatric Evaluation of Disability Inventory–Computer Adaptive Testing measure). We benchmarked this analysis with the cut points approach, calculated using thresholds from the literature which had been validated using the same device and for a population which most closely matched ours. Time spent active as measured by this unsupervised approach correlated more strongly with PEDI-CAT measures of the child's mobility (R 2 : 0.51 vs 0.39), social-cognitive capacity (R 2 : 0.32 vs 0.20), responsibility (R 2 : 0.21 vs 0.13), daily activity (R 2 : 0.35 vs 0.24), and age (R 2 : 0.15 vs 0.1) than that measured using the cut points approach. Unsupervised machine learning offers the potential to provide a more sensitive, appropriate, and cost-effective approach to quantifying physical activity behaviour in diverse populations, compared to the current cut points approach. This, in turn, supports research that is more inclusive of diverse or rapidly changing populations. Physical activity participation in young children has often been measured using parent reports. Accelerometry provides a more objective measurement but the traditional methods used to quantify this require calibration and struggle to generalise to diverse or rapidly changing populations such as young children. In recent years unsupervised machine learning methods have been shown to be able to segment and cluster accelerometry, allowing categories of activity intensity to emerge from a data-driven process. Here we show that an unsupervised machine learning technique (the hidden semi-Markov model) can be used to estimate categories of activity intensity in accelerometry data recorded from a diverse population of children age 9–36 months. We also show that this approach better captures the variance of movement abilities in the population than the traditional cut points approach. The hidden semi-Markov model approach provides a more effective approach for processing and analysing accelerometer data in rapidly changing and diverse populations such as young children, compared to the more traditional cut points approach. As it does not require calibration studies to incorporate new populations it has the potential to facilitate inclusion of unrepresented populations in research, as well as being less resource intensive.

The use of Latent Class Analysis (LCA) to Assess Children's Movement Behaviours Measured by Accelerometer and Self-report

Use of latent class analysis as a method of assessing the physical activity level, sedentary behavior and nutritional habit in the adolescents’ lifestyle: A scoping review

A systematic scoping review of latent class analysis applied to accelerometry-assessed physical activity and sedentary behavior

Age and School-Segment Difference in Daily Sedentary Behavior and Physical Activity among Student (9–23 Years): a Cross-Sectional Accelerometer-Based Survey

The Associations of Objectively Measured Physical Activity, Fundamental Motor Skills and Time in Sedentary Behavior in Children: A Cross-Sectional Study

School-Time Movement Behaviors and Fundamental Movement Skills in Preschoolers: An Isotemporal Reallocation Analysis

Twenty-four-hour movement behaviours and fundamental movement skills in preschool children: A compositional and isotemporal substitution analysis

Machine Learning Models for Classifying Physical Activity in Free-Living Preschool Children

Measuring Physical Activity in Children Who Do Not Walk and With Cerebral Palsy: An Exploratory Case Series

Multilevel analyses of school and children's characteristics associated with physical activity

Objective Measurement of 24-Hour Movement Behaviors in Preschool Children Using Wrist-Worn and Thigh-Worn Accelerometers

Physical Activity among U.S. Preschool-Aged Children: Application of Machine Learning Physical Activity Classification to the 2012 National Health and Nutrition Examination Survey National Youth Fitness Survey

Movement behaviours and physical, cognitive, and social-emotional development in preschool-aged children: Cross-sectional associations using compositional analyses

Individual, family, school and neighborhood predictors related to different levels of physical activity in adolescents: A cross-sectional study

Physical Education Classes, Physical Activity, and Sedentary Behavior in Children

International Comparison of the Levels and Potential Correlates of Objectively Measured Sedentary Time and Physical Activity among Three-to-Four-Year-Old Children

Using unsupervised machine learning to quantify physical activity from accelerometry in a diverse and rapidly changing population

The Cross-Sectional and Longitudinal Association Between 24-Hour Movement Behavior Compositions With Body Mass Index, Waist Circumference, and Internalizing and Externalizing Symptoms in 6-Year-Old Children

Comparison of self-report versus accelerometer – measured physical activity and sedentary behaviors and their association with body composition in Latin American countries

24-Hour Movement Guidelines: Descriptive Study with Overweight and Obese Low-Income Children

Temporal relationships in the movement behaviour of adolescents: Testing and methodological considerations of the ActivityStat hypothesis