Network meta-analysis of the effects of digital interventions on social functioning in children with autism spectrum...

This study performed a network meta-analysis of randomized and matched controlled trials to evaluate the effects of digital interventions on social functioning in children with autism spectrum disorder (ASD), compare the relative efficacy across digital modalities, and generate evidence to inform clinical decision-making and future research.

Following PRISMA-NMA guidelines, we systematically searched PubMed, Embase, Web of Science, Cochrane CENTRAL, and CINAHL for randomized or matched controlled trials evaluating digital interventions in children with ASD. The primary outcome was social functioning, and the secondary outcome was the dropout rate.A frequentist random-effects network meta-analysis was performed. Surface under the cumulative ranking curve (SUCRA) values were used to rank the interventions, and the certainty of evidence was assessed using the CINeMA framework.

Twenty trials involving 912 children with ASD were included in the analysis. The network meta-analysis indicated that all categories of digital interventions were generally well tolerated, with low dropout rates and no statistically significant differences between groups. For social functioning, computer-based intervention plus treatment as usual (CBI+TAU; SMD = 0.99, 95 % CI: -0.02 - 2.20), robot-based intervention plus treatment as usual (RBI+TAU; SMD = 0.72, 95 % CI: 0.15 - 1.30), and computer-based intervention (CBI; SMD = 0.84, 95 % CI: 0.44 - 1.24) demonstrated trends toward superiority over control conditions. CBI+TAU showed the largest point estimate and ranked highest in SUCRA; however, its effect estimate was imprecise and did not reach statistical significance. Although virtual reality-based interventions (VBI) and mobile-based interventions plus treatment as usual (MBI+TAU) ranked lower, both demonstrated meaningful clinical potential.

Digital interventions provide measurable benefits for improving social functioning in children with ASD and are generally well tolerated. CBI+TAU ranked highest in SUCRA; however, its effect estimate was imprecise and did not reach statistical significance, indicating substantial uncertainty. Therefore, SUCRA rankings should be interpreted as indicative of relative tendencies rather than definitive evidence of superiority. Despite limited direct evidence, VBI and MBI show considerable promise, particularly for home-based implementation. Clinically, CBI+TAU may be considered a promising option; however, its role as a core intervention should be interpreted cautiously until supported by larger, adequately powered trials. Other digital modalities, including RBI, MBI, and VBI, may serve as complementary components within multimodal intervention frameworks.