Artificial Intelligence Applications in Pediatric Dentistry: A Systematic Scoping Review With Visual Evidence Mapping

Artificial intelligence is increasingly applied to pediatric dental imaging, caries detection, orthodontic decision support, age estimation, and risk prediction. However, pediatric evidence remains fragmented across clinical tasks, model types, validation strategies, and reported metrics. This systematic scoping review mapped the clinical applications, methodological characteristics, and reported performance of AI in pediatric dentistry.

PubMed, Scopus, Web of Science, and Google Scholar were searched for English-language studies published from January 2010 to May 2026. Original studies evaluating AI, machine learning, or deep learning in pediatric dentistry or pediatric-relevant dental tasks were included. Screening was performed using Rayyan. Data extraction captured study characteristics, pediatric relevance, AI method, clinical task, outcomes, and validation approach. Risk of bias was assessed using QUADAS-3 for diagnostic/classification studies and PROBAST + AI for prediction and model-development studies. Findings were synthesized descriptively according to PRISMA-ScR.

Thirty-five studies were included, comprising 19 (54.3%) pediatric-only studies and 16 (45.7%) broader or mixed-age studies with separately identifiable pediatric or pediatric-relevant evidence. By primary application, 15 (42.9%) addressed orthodontic or craniofacial assessment and decision support, eight (22.9%) risk prediction or biomarker modeling, five (14.3%) diagnostic imaging or dental-anomaly detection, five (14.3%) dental age estimation, and two (5.7%) caries or plaque detection. Validation was internal or test-set based in 26 studies (74.3%), observer- or manual-comparative in five (14.3%), and not clearly defined in four (11.4%); none reported independent external validation. For QUADAS-3, non-low-risk judgments were most frequent in the analysis domain (8/19, 42.1%), whereas for PROBAST + AI they were most frequent in the participants and data-sources domain (5/16, 31.3%).

AI shows promising pediatric dental applications, especially in image-based and decision-support tasks, but evidence remains heterogeneous, task-specific, and limited by variable validation, incomplete reporting, and scarce external testing.