ADHD is typically assessed through reports from parents, teachers, or clinicians, but these reports may not fully capture how motor behavior is organized at the signal level. This cross-sectional study examined whether X-axis acceleration features derived from a wearable device could provide preliminary evidence of ADHD symptom-related motor-pattern differences in children and adolescents. Primary school children aged 6–13 years wore an Apple Watch Series 7, and X-axis accelerometer signals were used to extract features reflecting waveform distribution, zero-crossing rate, micro-motion, and local movement fragmentation. ADHD symptoms and broader emotional and behavioral difficulties were assessed using the SNAP-IV and SDQ. The results showed that the X-axis zero-crossing rate was the most robust feature differentiating the High ADHD and Low ADHD groups across the T-task and F-task recordings. X-axis zero-crossing rate reached statistical significance (p = 0.029), indicating more frequent short-interval switching between positive and negative acceleration directions in children with higher ADHD symptom levels. This finding suggests that directional switching or local movement fragmentation in X-axis acceleration may be a sensitive movement characteristic associated with ADHD symptoms. In addition, X-axis skewness showed a consistent directional tendency, with higher values in the High ADHD group; this effect was marginal in the T-task recording (p = 0.065), suggesting a possible tendency toward waveform asymmetry or distributional imbalance during longer movement recording. Overall, these findings provide preliminary evidence that ADHD symptom-related motor differences may be reflected in the organization of X-axis acceleration signals, particularly in directional switching indexed by zero-crossing rate and, to a lesser extent, waveform asymmetry indexed by skewness. Given the cross-sectional design, symptom-based grouping, modest sample size, and incomplete recording-context control, these results should be interpreted cautiously and require confirmation in larger, diagnostically characterized samples with standardized wearable-recording protocols.
Zhang et al. (Tue,) studied this question.