The role of artificial intelligence interventions to improve eye contact for children with autism spectrum disorder: a systematic review.

backgroundChildren with autism spectrum disorder (ASD) often struggle with eye contact during social interactions, a key aspect of effective communication. In recent years, various artificial intelligence (AI)-based interventions have been developed to support children with ASD. This study aims to systematically review the published literature on the AI-based interventions for improving eye contact.

methodsThe review adhered to the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Two researchers searched six databases: Scopus, Web of Science, PubMed, APA PsycInfo, Education Source, and IEEE Xplore, which initially yielded 1663 records. After screening and eligibility assessment, 16 studies met all inclusion criteria. Data were extracted through a narrative synthesis focusing on key variables, including study characteristics (author, year, region, sample, design, setting, and duration), definitions and measurements of eye contact, AI tool, and technological modalities, findings including reported effectiveness, generalization and maintenance outcomes, as well as whether social validity was assessed in each study. Two researchers independently conducted data extraction, quality assessment, and risk-of-bias evaluation using the Joanna Briggs Institute (JBI) checklists and the Single-Case Experimental Design (SCED) scale.

resultsIncluded studies used a variety of AI-based tools. The most frequently used intervention modality was robotic systems, which were used in 12 studies, followed by wearable technologies (two studies), virtual reality (one), and game-based software (one). Preliminary findings suggest that AI technologies are often associated with increases in the frequency and duration of eye contact. Six studies (38%) assessed generalization of treatment effects, generally reporting positive outcomes, while seven studies (44%) assessed social validity. The research mostly utilized quantitative designs, including randomized controlled trials, repeated measure designs, single-subject research designs, and quasi-experimental designs. One qualitative case study and one case report were also identified. Methodological quality ratings ranged from moderate to high across study designs.

conclusionThis systematic review revealed that AI-based interventions are developing technology for improving eye contact behavior in children with ASD. Particularly, robotic systems provide promising evidence for improving eye-contact behaviors in children with ASD. However, the current evidence base remains limited due to small sample sizes and variations in the operational definitions of eye contact, measurement methods, and levels of AI sophistication. Data on generalization and social validity were also limited, highlighting the need for larger, methodologically rigorous studies to confirm these findings and assess their broader applicability. CLINICAL TRIAL NUMBER: This study did not involve clinical trials, and as such, there is no clinical trial number to report.