Evaluating the Comparative Effectiveness of Traditional and AI-Assisted Speech Therapy on Communication Outcomes in Children with Autism Spectrum Disorder

Abstract

Communication impairments are major challenges for children with Autism Spectrum Disorder (ASD), affecting expressive language, receptive language, social interaction, and adaptive functioning. The growing need for accessible and individualized speech-language interventions has increased interest in artificial intelligence (AI)-assisted technologies as complementary clinical tools. This study compared the effectiveness of Traditional Speech Therapy (TST) and AI-Assisted Speech Therapy (AIST) in improving communication outcomes among children with ASD in Pakistan. A randomized controlled trial was conducted with 100 children aged 5–12 years. Participants were selected through purposive sampling from selected special education schools, autism rehabilitation centers, and speech therapy clinics, and were then randomly assigned to the TST group (n = 50) or AIST group (n = 50). Both groups received three 45-minute sessions per week for 12 weeks. Outcomes were assessed using the CCC-2, Vineland Communication Domain, ATEC Communication Subscale, and Therapist Observation Checklist. The AIST group showed significantly greater improvements in expressive and receptive language, social communication, eye contact, verbal initiation, turn-taking, and overall communication performance (p < .001). Large effect sizes (Cohen’s d = 1.14–1.46) and ANCOVA findings confirmed the superiority of AIST after controlling for baseline ability. The findings support AI-assisted speech therapy as an evidence-based complement to conventional rehabilitation.