Can motion sensors accurately measure postures and movements among children?

The need for accurate measurements

One of the common concerns of parents and professionals related to screen use is that it is mainly performed while sitting and that this may displace physical activity. Strong evidence shows that physical activity is essential for children’s physical and mental wellbeing. However, the link between screen use and physical activity is unclear. Some evidence suggests that high amounts of screen use are related to lower levels of physical activity, whileother evidence suggests that children can engage in high amounts of screen use as well asphysical activity.

One of the limitations of the current evidence is the poor precision of measures of children’s screen use and their postures and movements when engaged with screens. We therefore aimed to develop more accurate methods for capturing children’s postures and movements.

Using motion sensors to capture postures and movements

We had 48 children (aged 3-14 years) do a series of activities in a laboratory for around an hour while being video recorded. The activities included common postures and movements, i.e. lying, sitting, standing, walking, running and stair climbing. The children wore onemotion sensor (accelerometer) attached to the front of their right thigh. We processed the sensor data using two algorithms developed to identify the postures and movements of adults. We compared the postures and movements estimated by the algorithms with the video recordings which researchers had coded for the activities the children were doing. 

We found that both algorithms could estimate a range of children’s postures and movements with reasonable accuracy. Specifically, the algorithms performed well in estimating time spent lying, sitting, and standing. However, lower accuracies were found for classifyingwalking, running and stair climbing, particularly among younger children. The lower performance among younger children is likely a result of their movement patterns being more different than adults which the algorithms were originally developed for. 

In conclusion, sensors worn on children’s thighs can provide important insight into children’s daily postures and movements with very little burden on children or their families. However, further work is required to improve the methods’ accuracy in capturing children’s time spend in more dynamic movements, particularly among younger children. Such methods could be used to better investigate how children’s screen use might influence physical activity and sitting.