Validation of an accelerometer system for activity monitoring in children with functional disabilities

To assess the validity of the SENS motion® system (SENS) for measuring postures and movements in school-aged children with functional disabilities, using direct video observation as the criterion method. In this cross-sectional-study, 29 children (51.7% male, mean age 10.8 ± 2.9 years) from two special schools in Denmark participated. Each child wore a SENS device on the thigh while completing a standardized protocol of six categories: lying/sitting, standing, walking, running, cycling, and step count. All activities were video recorded. Video data were coded in 5-s epochs and aligned with SENS output. Agreement was assessed by comparing the observed time spent in each activity with the corresponding time estimated by SENS, expressed as mean values, standard deviations, mean differences, and percentage agreement. SENS showed excellent agreement for walking (93.2%) and lying/sitting (96.2%), good agreement for running (89.2%), and moderate agreement for standing (74.1%). Cycling was poorly detected, with only 6.4% agreement, as it was frequently misclassified as walking (50.8%) or lying/sitting (42.7%). Step counts were slightly overestimated by SENS (mean difference 7.2 steps). Overall, SENS tended to underestimate activity duration compared with the observation.

SENS demonstrated excellent to good validity for detecting lying/sitting, walking, and acceptable validity for running in children with functional disabilities, but moderate validity for standing and limited validity for cycling. These findings indicate that SENS may be useful for monitoring several common postures and activities in this population, though further algorithm refinement and broader validation are needed, particularly for cycling and postural transitions.

• Children with disabilities engage in less dynamic postures and more lying/sitting time than their peers without disabilities. • Accelerometer systems are typically validated in healthy populations, limiting accuracy in children with disabilities.

• SENS shows excellent to good validity for lying/sitting, walking, and running in children with functional disabilities. • SENS performs poorly for cycling and moderately for standing, highlighting the need for algorithm refinement.