Comparative Investigation of Smartwatch Data in Children with ADHD and Non-ADHD

This manuscript investigates behavioral differences between children with ADHD and neurotypical children (6-11y) using smartwatch data from both hands collected during four classroom-like activities. Our study analyzes accelerometer and gyroscope data to identify distinct behavioral patterns, that can be used to enhance diagnostic methods for ADHD. We pre-processed the smartwatch data and calculated statistical features such as mean, median, standard deviation, kurtosis, and sum of consecutive differences. Wrist angles were used to differentiate on-task and off-task states. Hypothesis testing indicated that certain activities and features, such as the calculated angle of the y-axis in activity 1 (magnetic tiles) and the 'sum of consecutive differences' in activity 4 (drawing), were significant in distinguishing between the two groups. The left hand (non-dominant hand) generally provided more discriminative data than the right hand (dominant hand). Our findings suggest that specific patterns in smartwatch data can differentiate between ADHD and neurotypical children during classroom activities. These insights hold promise for developing more objective and data-driven diagnostic tools for ADHD.