Association Between Task Difficulty Regulation in Game-Based Digital Therapeutics and Attention-Deficit/Hyperactivity Disorder Symptoms in Children: Secondary Analysis of a Randomized Controlled Trial.

Background: Recent advances in digital therapeutics (DTx) have led to the development of game-based, home-delivered interventions for children with attention-deficit/hyperactivity disorder (ADHD). Prior studies have demonstrated the feasibility and clinical potential of DTx for reducing ADHD symptoms but have primarily evaluated usage metrics, including training frequency and duration, which do not capture in-game performance or learning processes during training. In addition, the association between task difficulty regulation during DTx training and ADHD symptom improvement in children has not been well studied. Objective: This study aimed to examine whether task difficulty regulation during DTx training is associated with ADHD symptom improvement in children and to evaluate the clinical relevance of difficulty-related behavioral measures as indicators of treatment response. Methods: In this study, a secondary post hoc analysis of data from 35 children aged 6 to 13 years with ADHD who participated in a randomized controlled trial was performed. The participants had completed a 4-week training program comprising 5 digital cognitive tasks from the DTx Neuro-World. During training, task difficulty was automatically adjusted in real time according to participants' response accuracy to maintain an appropriate challenge level. Training difficulty was quantified by 3 indices: average task difficulty, task difficulty variability, and task difficulty slope. Associations between difficulty indicators and ADHD symptom changes were examined using regression analyses and group comparisons. Results: We found that lower average task difficulty during tasks 1, 2, and 4 of Neuro-World was associated with improvements in ADHD symptoms, as assessed by the total Korean-ADHD Rating Scale (K-ARS) score. In addition, the average task difficulty of tasks 1 to 4 was related to improvements in hyperactivity/impulsivity symptoms of ADHD, as assessed by the K-ARS hyperactivity or impulsivity subscore, but not to improvements in inattention symptoms. Moreover, lower variability of task difficulty in task 5 improved ADHD hyperactivity/impulsivity symptoms, as assessed by K-ARS, but did not affect Korean-Child Behavior Checklist scores. Finally, we found that a negative task difficulty slope was associated with the alleviation of ADHD hyperactivity or impulsivity. These findings should be interpreted with caution, given the exploratory nature of the study and the absence of formal correction for multiple comparisons. Conclusions: As an exploratory investigation, our study provides a novel perspective by demonstrating that individualized and stable regulation of task difficulty, rather than the mere attainment of higher difficulty levels, is associated with ADHD symptom improvement. These findings indicate that dynamic difficulty trajectories derived from in-game behavioral data are clinically relevant, going beyond the conventional focus in the literature on usage metrics or overall training intensity. Our results support the role of individualized difficulty trajectories as potential digital biomarkers of clinical response. This approach has important implications for the design of DTx that incorporate personalized and adaptive difficulty regulation to enhance clinical effectiveness and scalability.