Conversational Artificial Intelligence in Pediatric Occupational Therapy: A Comprehensive Narrative Review of Emerging Applications, Transferable Evidence, and Future Directions

Abstract

Background: Conversational artificial intelligence (AI), including chatbots, large language models, virtual agents, and embodied conversational systems, is increasingly being introduced into healthcare, rehabilitation, education, and professional practice. In pediatric occupational therapy, these technologies could potentially support clinical documentation, professional reasoning, parent coaching, therapeutic engagement, home-program implementation, and access to information. However, the profession-specific evidence remains fragmented, and the implications for children, families, and occupational therapists have not been comprehensively examined. Objective: This comprehensive narrative review aims to critically examine the current and emerging applications of conversational AI in pediatric occupational therapy, integrate direct and transferable evidence, evaluate its alignment with occupation- and family-centered practice, and identify priorities for responsible future development. Methods: A structured multidisciplinary literature search was conducted across health, rehabilitation, allied-health, education, and computer-science databases. The narrative synthesis was guided by the quality domains of the Scale for the Assessment of Narrative Review Articles. Evidence was organized across four interconnected levels: direct pediatric occupational therapy research, broader occupational therapy applications, transferable pediatric evidence, and foundational literature concerning conversational-AI design, implementation, and governance. Results: Direct pediatric occupational therapy evidence remains limited but demonstrates two emerging directions: therapist-facing large language models supporting clinical documentation and child-facing embodied conversational systems intended to facilitate therapeutic engagement. The wider literature indicates potential applications in clinical reasoning, professional education, reflective practice, caregiver coaching, adherence, home routines, waiting-list support, and personalized communication. Nevertheless, most evidence is preliminary, heterogeneous, and based on small samples, technical prototypes, simulations, or indirect pediatric applications. Major unresolved concerns include developmental appropriateness, hallucinated information, algorithmic bias, privacy, safeguarding, emotional attachment, automation bias, professional accountability, and limited evaluation of participation and occupational outcomes. Conclusions: Conversational AI may become a valuable augmentative tool in pediatric occupational therapy, but it should not replace occupational therapists, contextual clinical reasoning, or therapeutic relationships. Future systems should be occupation-centered, family-centered, developmentally appropriate, transparent, context-sensitive, and governed through professional oversight. Co-design with children, caregivers, and occupational therapists, together with rigorous real-world evaluation, will be important before routine clinical use can be considered.

The authors' abstract, as published at the source. International Medical Education, 2026 · DOI ↗