Differentiating Pediatric Bipolar Disorder and ADHD Using Actigraphy and Self-Reported Mood Data with Machine Learning Prediction

This study examined 209 inpatients to distinguish bipolar disorder from ADHD by analyzing objective activity patterns (actigraphy) alongside self-reported mood and energy ratings. Machine learning models using actigraphy, sleep, and demographic data successfully predicted severe mood episodes on the same day and the following day, with bipolar disorder showing tighter coupling between activity extremes and mood disturbance.