Comparative Effectiveness of Artificial Intelligence-Based Interventions in the Treatment of Reading Disorders (Dyslexia) in Elementary School Students: A Meta-Analysis (۲۰۲۰-۲۰۲۵)

This meta-analysis systematically evaluates the efficacy of Artificial Intelligence (AI)-based interventions for elementary school students diagnosed with dyslexia. Adhering to PRISMA guidelines, a comprehensive search was conducted across prominent global academic databases, including Google Scholar, PubMed, Scopus, ERIC, and key Iranian scientific repositories such as SID and Civilica. The search focused on peer-reviewed studies published between January ۲۰۲۰ and December ۲۰۲۵. A rigorous screening process identified ۱۲ high-quality studies, encompassing a total of ۲,۵۳۰ participants, that met the predefined inclusion criteria. The analysis primarily investigated two distinct yet complementary dimensions of AI application: AI-based pedagogical interventions designed to enhance reading skills, and AI-driven detection and prediction systems aimed at early and accurate diagnosis. The aggregated results indicated a small but statistically significant positive overall effect size (d = ۰.۱۶, ۹۵% CI [-۰.۲۰, ۰.۵۲]) for AI-based pedagogical interventions, though this was accompanied by substantial heterogeneity (I² = ۹۵.۴۸%), suggesting considerable variability across studies. Further subgroup analyses revealed that interventions incorporating personalized intelligent tutoring systems, adaptive learning platforms, and speech-recognition tools generally demonstrated higher efficacy compared to more generalized AI software. Conversely, AI-driven detection and prediction systems consistently exhibited high accuracy, offering promising avenues for early identification. The findings underscore AI's transformative potential in both early detection and personalized educational support for dyslexic students. However, the observed heterogeneity highlights the critical need for standardized intervention protocols and further research into contextual factors influencing effectiveness. This meta-analysis concludes that while AI is a powerful adjunct to traditional methods, its optimal integration into diverse educational settings requires careful consideration of specific student needs and intervention design.

کلیدواژه ها:

Dyslexia ، Artificial Intelligence ، Elementary Education ، Meta-Analysis ، Reading Disorders ، Machine Learning ، Pedagogical Interventions ، Early Detection ، Personalized Learning ، Educational Technology