JMIR Publications

ABSTRACT

Background:

Mobile game–based cognitive training may improve untrained task performance in children with attention-deficit/hyperactivity disorder (ADHD). However, aggregate task scores cannot show whether improvement is uniform across a task's component operations, which eye-movement tasks allow to be scored separately.

Objective:

This study aimed to determine whether differential longitudinal change after a 4-week mobile cognitive training program was uniform across observable components of oculomotor control in children with ADHD compared with healthy controls (HCs).

Methods:

In this secondary analysis of a longitudinal feasibility study, 110 children aged 6 to 12 years (55 with ADHD and 55 HCs) completed antisaccade (AS) and memory-guided saccade (MGS) tasks before and after the same 4-week home-based mobile serious game training program. Outcomes were AS directional control and 3 MGS components: successful gating, conditional direction, and conditional spatial fidelity at 1-, 3-, and 5-second delays. Differential change (ADHD vs HC) was estimated as a difference-in-differences (DiD) using mixed models with participant-cluster bootstrap inference and Holm correction; the 3 binary components were also compared directly in an exploratory analysis.

Results:

Of the 110 participants, 103 (51 ADHD and 52 HC) had a valid post-assessment. AS directional control increased from 0.146 to 0.236 in the ADHD group and decreased from 0.189 to 0.173 in the HC group (DiD 0.106, 95% CI 0.033-0.179; Holm-adjusted P=.02). Successful gating showed a similar but less precise effect (DiD 0.089, 95% CI −0.029 to 0.207; Holm-adjusted P=.41), whereas conditional direction did not (DiD −0.031, 95% CI −0.084 to 0.021; Holm-adjusted P=.49). Conditional spatial deviation decreased in both groups without differential change (ratio of ratios 1.001, 95% CI 0.719-1.393; Holm-adjusted P>.99). The exploratory direct comparison indicated heterogeneity among the binary effects (χ22=11.95; P=.003): the AS effect exceeded the conditional direction effect by 0.137 (95% CI 0.055-0.219; Holm-adjusted P=.003), whereas its difference from the gating effect was small and imprecise (0.016, 95% CI −0.122 to 0.154). No retention-interval contrast and no association with Continuous Performance Test or parent-rated executive function outcomes survived multiplicity correction.

Conclusions:

Differential change after mobile cognitive training in children with ADHD was not uniform across oculomotor components: AS directional control changed more than post-go direction selection in the MGS task. Because both groups trained and there was no untreated ADHD comparator, these findings describe differential longitudinal change rather than treatment efficacy. Component-level scoring can reveal patterns hidden by aggregate task scores and may inform endpoint selection in randomized controlled trials of digital interventions. Clinical Trial: ClinicalTrials.gov NCT06181747; https://clinicaltrials.gov/study/NCT06181747