A systematic review of 13 studies covering 25 years of research finds that child-centered play therapy improves social interaction, communication, emotional regulation, and repetitive behaviors in children with autism spectrum disorder—and points to AI and virtual reality as the next frontier for personalized intervention.
As autism spectrum disorder (ASD) affects millions of children worldwide, researchers continue to search for interventions that respect children’s natural modes of expression while delivering measurable therapeutic benefits. A new scoping review published in Frontiers of Digital Education provides a comprehensive synthesis of evidence on child-centered play therapy (CCPT)—a humanistic approach that positions children as active agents in their own healing—and outlines how intelligent digital technologies could address the field's longstanding limitations.
The study, conducted by Jingying Chen, Suyun Tang, and Guangshuai Wang from the National Engineering Research Center for Educational Big Data at Central China Normal University, was published on September 10, 2026.
25 Years of Evidence: Consistent but Heterogeneous Outcomes
The researchers analyzed 13 studies published between 2000 and 2025, involving 130 children with ASD aged 3 to 11. Following PRISMA-ScR guidelines, the review examined intervention protocols, outcome measures, and therapeutic effects across a wide range of settings—from university clinics and school playrooms to rural classrooms.
The findings indicate that CCPT, which is grounded in humanistic psychology and emphasizes children’s innate potential for self-healing and growth, produces measurable improvements across multiple domains. Participation in CCPT was associated with enhancements in social interaction (12 studies), communication difficulties (6 studies), repetitive behaviors (6 studies), baseline behavioral functioning (4 studies), emotional issues (4 studies), and adverse childhood experiences (1 study).
Notably, the intervention group in one randomized controlled trial showed a significant decrease in Social Responsiveness Scale-2 total scores (78.83 to 70.58), while the control group’s symptoms worsened (73.64 to 77.91), yielding a large effect size (ηp² = 0.47). Another study found that CCPT significantly boosted alpha wave activity (p < 0.001) as measured by electroencephalography (EEG), indicating improved neural relaxation and an enhanced ability to suppress distracting information in social situations.
However, the review also revealed substantial individual differences. High-functioning children demonstrated more pronounced improvements in social skills and empathy, whereas low-functioning children required extended periods to establish therapeutic rapport. Some studies reported mixed effects, with certain participants showing no improvement or even regression in specific domains.
From Playroom to Daily Life: The Generalization Challenge
While CCPT’s flexibility allows it to be combined with other approaches—including robot-assisted play, music therapy, and rhythmic synchronization activities—the review identified a persistent limitation: Skills acquired in the playroom often show limited transferability to children’s daily lives.
“VR and augmented reality technologies address this gap by simulating authentic social scenarios, effectively bridging the divide between therapeutic acquisition and real-world application,” the authors note. Virtual playrooms eliminate constraints imposed by physical time and space, allowing for feasible adjustments based on children’s individual needs while mitigating concerns about fatigue and emotional fluctuations affecting therapeutic outcomes.
A New Paradigm: “Child-Led + Data-Driven” CCPT
The review proposes a forward-looking theoretical framework for what the authors call a “child-led + data-driven” two-dimensional CCPT paradigm. In this model, children maintain autonomy in selecting physical or virtual toys, determining play themes, and freely transitioning between VR and AR environments, while therapists adhere strictly to nondirective principles.
Simultaneously, multimodal data collection—including behavioral data (play duration, toy selection frequency), physiological data (eye-tracking, heart rate variability, galvanic skin response), and affective computing (facial micro-expressions, vocal affect)—provides objective therapeutic benchmarks. Machine learning identifies deviations as “critical behavioral nodes,” enabling dynamic adjustments to the therapeutic environment.
The authors emphasize that these two dimensions are not mutually exclusive but interact dynamically through intelligent systems. “Data systems provide indirect guidance through environmental fine-tuning rather than direct instruction,” they write. When AI detects an increase in a child’s anxiety—evidenced by repeated discarding of virtual toys—the VR scenario might automatically dim the lighting, play soothing music, or summon a docile virtual pet. Yet the ultimate decision regarding whether and how to intervene remains with the therapist, avoiding overreliance on technology.
Addressing the Evidence Gap
The review also highlights a critical shortage of high-quality studies. Among the 13 reviewed studies, only 2 were randomized controlled trials meeting high-quality criteria. The authors attribute this evidence gap to four methodological challenges: the inherent complexity of CCPT’s relational nature, the heterogeneity of the ASD population, resource constraints, and the inadequate use of digital tools in existing research.
“The predominance of case studies may reflect that the effect of CCPT in educational interventions for children with ASD remains at an exploratory stage,” the authors note. “Future research should integrate both paradigms: Case studies may inform localized variables for randomized controlled trials (RCTs), while rigorous RCTs could help correct over-interpreted associations from case studies.”
Implications for Digital Educational Tool Design
The findings offer three key guidelines for designing digital educational tools and learning environments for children with ASD. First, the significant individual differences revealed by the research necessitate tools with deep personalization capabilities. Second, digital environments should prioritize constructing open-ended interactive narratives and game scenarios that foster social initiative, joint attention, and symbolic play, rather than isolated skills-training modules. Third, a blended learning ecosystem—seamlessly integrating quantifiable, adaptable personalized digital practice with opportunities for skill application in safe, supportive real or virtual social environments—would create a complete closed loop from personalized training to socialized application.
This design philosophy—combining a “data-driven personalized core” with a “child-led interactive shell”—constitutes the systemic blueprint for implementing the proposed paradigm.
DOI: 10.1007/s44366-026-0098-7