Abstract
OBJECTIVE: Attention-deficit/hyperactivity disorder (ADHD) is a prevalent neurodevelopmental disorder in children and adolescents. Diagnoses of ADHD often rely on subjective ratings from parents and teachers. This study investigated the feasibility of using objective activity monitoring data collected through wearable activity trackers for ADHD diagnosis and monitoring in adolescents. METHOD: A longitudinal study was conducted involving Chinese adolescents ages 16 to 17 years. Data collected included objective measures (movement acceleration, heart rate, and sleep patterns from passive actigraphy) and subjective measures (parent and self-reported questionnaires). Machine learning models were developed using eXtreme Gradient Boosting (XGBoost) to compare various measures for ADHD classification. Model performance was evaluated using the area under the receiver operating characteristics curve (AUC). The SHapley Additive exPlanations (SHAP) method was used to analyze the importance of different measures on ADHD risk. RESULTS: The study included 30 adolescents (17 with ADHD and 13 without ADHD). Machine learning models using solely objective measures achieved high predictability in classifying ADHD (AUC = 0.844) and ADHD medication status (AUC = 1.000). Models integrating both subjective and objective measures showed enhanced performance (AUC = 0.933). In this sample, key features for ADHD classification included irritability, sex, and quality-of-life indicators; key features for ADHD medication use classification included heart rate and physical activity intensity. CONCLUSION: Although the sample size was small, actigraphy-based monitoring provides a noninvasive and granular measurement of objective vital signs of adolescents. If validated in larger samples, the incorporation of objective measures is likely to enhance multidimensional assessment and diagnostic accuracy in adolescents with ADHD, supplementing existing diagnostic methods. STUDY PREREGISTRATION INFORMATION: A systematic review of anhedonia and amotivation in depression and cannabis use; https://www.crd.york.ac.uk/prospero/; CRD42023422438.