Characterizing infant leg movements using 72-h wearable sensor data: Descriptive analysis from a large, heterogenous...

Spontaneous limb movements provide the foundation for motor development, yet knowledge of their normative characteristics has been limited by small homogenous samples and short observation windows. Leveraging the HEALthy Brain and Child Development (HBCD) study, we present descriptive characteristics of the first large-scale dataset of leg movements from 421 infants aged 0-2 months, captured with wearable sensors worn continuously for 72 h in naturalistic settings across multiple research sites in the United States. Our analysis focused on three domains: movement characteristics, variability of movement acceleration time-series, and physical activity intensity. Movement characteristics included leg movements per hour awake, peak acceleration per movement, average acceleration per movement, and movement duration. These characteristics aligned with earlier smaller-scale studies and showed highly consistent patterns between right and left legs. Sample entropy analysis revealed left-skewed distributions with median values near 1.3. Physical activity intensity estimations showed that infants spent the majority of time in sedentary activity, followed by light activity, with only brief periods of moderate-to-vigorous activity. Together, these findings provide initial characterization of multiple dimensions of infant leg movements, hence significantly contributing to derive reference distributions in early life. By establishing scalable, ecologically valid, and computationally tractable measures of infant motor behavior, this study lays the groundwork for integrating wearable sensing with longitudinal developmental science and for identifying early indicators of atypical trajectories. Ultimately, these findings contribute to the broader goals of the HBCD study: to understand how the brain develops and is shaped by environmental, social, and biological factors during pregnancy and after birth.