ARTIFICIAL INTELLIGENCE IN DENTAL CARIES DIAGNOSIS IN PEDIATRIC DENTISTRY: AN INTEGRATIVE LITERATURE REVIEW

Abstract

Introduction: Early childhood caries represents a persistent public health challenge, in which early diagnosis in primary teeth is hindered by limited pediatric cooperation, overlapping anatomical structures, and the subjectivity of conventional clinical-radiographic examination. In this context, artificial intelligence (AI) emerges as a promising decision-support tool. Objective: To evaluate the diagnostic accuracy and clinical feasibility of artificial intelligence in detecting carious lesions in primary and mixed dentition, as well as its role in screening and routine pediatric dentistry practice. Methodology: An integrative literature review was conducted using the PubMed/MEDLINE, LILACS, and Google Scholar databases. The search encompassed articles published between 2025 and 2026 in Portuguese and English. Twelve studies were selected, including systematic reviews, validation trials, and observational research addressing the application of AI models to intraoral radiographs and photographs. Results and discussion: The literature demonstrates that artificial intelligence exhibits high diagnostic accuracy, with sensitivity and specificity values frequently exceeding 85%. Computer vision models and convolutional neural networks stood out in identifying occlusal and proximal dentin caries, demonstrating performance comparable to experienced pediatric dentists and reducing inter-observer discrepancies. The use of AI on smartphone-captured photographs enabled population screening in Basic Health Units and schools. However, limitations remain regarding incipient enamel lesions and image contamination interference. Conclusion: Artificial intelligence constitutes an effective adjuvant tool for caries diagnosis and screening in pediatric dentistry. Its application functions as a second opinion system and should be combined with thorough clinical examination and the practitioner's professional judgment.