Can technology-based interventions be an alternative method for emotion recognition in autism spectrum disorder: Three-level meta-analysis insights from RCTs
- Forfatter(e)
- Xiao, H. L., Zhu, H., Liu, Z. Y., Zeng, T. A., Wang, J. Y., Shataer, S., Wang, H. P., Yang, C. J.
- År
- 2026
- Tidsskrift
- Research in Autism
- Volum
- 133
- Sider
- 14
- Kategori(er)
- Autismespekter Sosiale ferdigheter (inkl. vennerelasjoner)
- Tiltakstype(r)
- E-helsetiltak (spill, internett, telefon)
- Abstract
Background: Technology-based interventions (TBIs) have shown promise for improving emotion recognition (ER) in individuals with autism spectrum disorder (ASD). but the durability of effects and the role of technological features remain unclear. This meta-analysis examined short-term and maintenance outcomes and tested age, technological interactivity, and their interaction as moderators.
Methods: A systematic literature search of PubMed, MEDLINE, APA PsycINFO, ERIC, IEEE Xplore, and Web of Science was performed from database inception to May 31, 2024. Randomized controlled trials (RCTs) evaluating TBIs for ER in individuals with ASD were included. Twentytwo studies with 893 participants met the inclusion criteria. Effect sizes (Hedges' g) were pooled using a three-level meta-analytic model.
Results: TBIs were associated with a moderate short-term improvement in ER (g = 0.68), whereas the follow-up effect was smaller but not robust (g = 0.35). Moderator analyses did not identify statistically significant effects of age, technological interactivity, or their interaction. Statistically significant effects were observed in younger children (3-6 years: g = 1.17; 7-12 years: g = 0.63), and in lower-interactivity interventions (low: g = 0.93; moderate: g = 0.69).
Conclusions: TBIs may yield short-term improvements in ER in individuals with ASD, but evidence for sustained effects remains limited. Age, technological interactivity, and their interaction were not robust moderators, despite descriptively larger effects in younger children and lower-interactivity interventions. Given the substantial heterogeneity and uneven sample distribution, larger and more balanced studies are needed to confirm effect durability and identify reliable moderators.