Evaluating the effectiveness of intelligent interaction technology in autism interventions: A meta-analysis based on trial assessment.
Virtual and augmented reality give a reliable medium boost for autistic kids, especially preschoolers, while robot tools need tighter programming and adult pairing.
01Research in Context
What this study did
Zhou et al. (2025) pooled 13 randomized trials that tested smart tech for kids with autism. The tech included virtual-reality headsets, mixed-reality games, and small humanoid robots.
Together the trials covered 459 children. The team asked: do these gadgets beat control conditions on social, thinking, or behavior goals?
What they found
XR tools—think VR headsets and AR story tables—gave a clear medium boost. The overall effect size was 0.80, a jump you can feel in one semester.
Robot tools also helped, but the gains were smaller and less steady. Preschoolers gained the most across all tech types.
How this fits with other research
Rojahn et al. (2012) warned that robot studies were "exploratory and weak." Quan’s new pile of 13 RCTs now shows those early fears were partly right—robots work, just not as well as XR.
Kostrubiec et al. (2020) ran a single study where robot praise beat human praise for keeping low-functioning kids glued to the task, yet social gains faded without a human mediator. That lines up with Quan: robots engage, but need adult backup.
Soleiman et al. (2023) got strong emotion-recognition scores with two robots in an ABAB design. Quan’s meta-average smooths out such flashy single cases and tells us to expect milder, real-world robot effects.
Why it matters
If you serve preschoolers, XR deserves a front-row seat in your toolkit—medium effects mean faster IEP goal mastery. For robots, use them as spice, not the main dish: keep the human therapist in the loop and measure social carry-over weekly. Either way, write tech trials into your behavior plan and track generalization beyond the screen or bot.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Book a 10-minute XR emotion-recognition game this week, take baseline probe data before and after, and compare to your last table-top lesson.
02At a glance
03Original abstract
PURPOSE: This research aims to conduct a systematic review of the literature on the utilization of intelligent interaction technologies, including Extended Reality (XR) and robotic systems, within the context of autism interventions. METHOD: This study commenced on June 16, 2022, and conducted a systematic search of publications from 2017 to 2024 using Boolean terms such as "Virtual Reality" AND "autism" AND "RCT" in PubMed, Web of Science, Cochrane Library, SpringerLink, and Embase. Two reviewers independently conducted research selection, data extraction, and quality assessment (using Cochrane Risk of Bias tool, RoB 2), and resolved differences through discussion. Statistical analysis was conducted in STATA using a random-effects model to calculate Standardized Mean Difference (SMD), supplemented by sensitivity analysis, funnel plot, and Egger's test to ensure robustness. RESULT: The meta-analysis included 13 studies involving 459 individuals with ASD from different regions (age range: 2-15 years, diagnosed using standardized instruments), including Europe, the Americas, Asia, and the Middle East, but without representatives from Africa. The results showed that intelligent interactive intervention demonstrated significant efficacy (SMD=0.66, 95 % CI: 0.27-1.05, p < 0.001). Subgroup analyses indicated that age and intervention targets influenced the effect size, with particularly pronounced positive impacts observed in preschool-aged children and cognitive interventions. Subgroup analysis of intervention measures showed that the XR group exhibited a more positive effect. In contrast, the robotic group, due to high heterogeneity and wide confidence intervals, did not lead to a conclusion that this intervention was effective overall. Additionally, sensitivity analysis, funnel plots, and Egger's test were conducted, demonstrating that the results are stable and no significant publication bias. CONCLUSION: In conclusion, the overall efficacy of intelligent interaction as an intervention is positive, showing benefits in social, cognitive, and behavioral aspects of ASD, particularly with XR (SMD=0.80, 95 % CI: 0.47-1.13). However, further research is required to ascertain the effects of robots. Furthermore, this study suggests intelligent interaction demonstrated larger effects for preschool-aged children(2-6 years; SMD=1.00, p = 0.007).
Research in developmental disabilities, 2025 · doi:10.1016/j.ridd.2025.105087