Artificial Intelligence Chatbots and Large Language Models for Traumatic Dental Injury Management and Decision Support: A Systematic Review With Function-Based Narrative Synthesis and Exploratory Meta-Analysis.

Traumatic dental injuries require rapid, accurate, and context-sensitive decisions, particularly during emergencies such as tooth avulsion, luxation injuries, and trauma involving the primary dentition. Artificial intelligence (AI) chatbots and large language models (LLMs) are increasingly used by patients, caregivers, students, and clinicians for immediate information and decision support, yet their reliability in dental trauma remains uncertain. This systematic review aimed to evaluate the accuracy, consistency, safety, guideline concordance, and clinical applicability of AI chatbots and LLMs for traumatic dental injury management and decision support. PubMed, Scopus, and Embase were searched through April 2026. Eligible studies evaluated AI-generated responses against clinical guidelines, expert assessment, validated dental trauma scenarios, photographs, real cases, standardized questions, or other reproducible reference standards. Twenty-eight studies met the eligibility criteria. A function-based narrative synthesis classified the evidence into five translational functions: patient- and caregiver-facing emergency guidance, guideline-based diagnostic and management benchmarking, domain-specific or document-grounded workflows, multimodal and real-case validation, and avulsion-specific prognostic or advanced decision support. An exploratory meta-analysis of nine study-level estimates with comparable binary outcomes demonstrated a pooled proportion of accurate or guideline-concordant responses of 81.7%, although heterogeneity was very high. Across studies, performance varied according to model, prompt format, reference standard, input type, and assessment methodology. Safety concerns included incomplete recommendations, misleading information, inconsistent responses, limited readability, unreliable references, and injury-specific performance variability. Overall, current evidence indicates that AI chatbots and LLMs can support dental trauma information retrieval and supervised clinical decision support, particularly when constrained by curated guidelines or domain-specific workflows. However, the available evidence remains limited by substantial methodological heterogeneity and limited real-world validation and therefore does not support the autonomous use of these systems for emergency traumatic dental injury management. Trial Registration: PROSPERO; CRD420261394046.