What question did this study set out to answer?

This review aims to assess and synthesize the clinical applications and diagnostic performance of AI and ML in various dental specialties.

January 18, 2026Open Access

Artificial Intelligence and Machine Learning in Dentistry: Evidence-Based Perspectives — A Systematic Review

Key Points

This review aims to assess and synthesize the clinical applications and diagnostic performance of AI and ML in various dental specialties.
Conducted a systematic search following PRISMA 2020 guidelines across multiple databases.
Included English-language studies from January 2019 to December 2024 that utilized AI or ML in dentistry.
Applied the QUADAS-2 tool for quality assessment and calculated pooled sensitivity, specificity, and accuracy values.
39 studies met inclusion criteria from an initial 342 identified.
AI models achieved a pooled diagnostic accuracy of 92.4%, with sensitivity at 90.1% and specificity at 91.7%.
AI outperformed human experts in 68% of comparative studies, with significant improvements in diagnostic sensitivity for image-based tasks.

Abstract

Abstract Background: Artificial Intelligence (AI) and Machine Learning (ML) are reshaping modern dentistry by enhancing diagnostic precision, optimizing treatment planning, and improving patient management. These data-driven systems can analyze vast clinical datasets, uncover hidden patterns, and provide evidence-based clinical support. Objective: To systematically evaluate and analyze the applications, diagnostic accuracy, and outcome measures of AI and ML tools across various dental specialties, and to synthesize statistical evidence supporting their clinical performance. Methods: Following the PRISMA 2020 guidelines, a comprehensive search was performed across PubMed, Scopus, Web of Science, and Google Scholar for studies published between January 2019 and December 2024. Inclusion criteria comprised original English-language studies employing AI or ML for diagnostic, predictive, or prognostic purposes in dentistry. The QUADAS-2 tool was applied for quality and bias assessment. Pooled sensitivity, specificity, and accuracy values were computed for AI-based diagnostic models using a weighted mean analysis. Results: Of 342 studies initially identified, 39 met inclusion criteria. The pooled diagnostic accuracy of AI models was 92.4% (95% CI: 88.6–96.3%), sensitivity 90.1% (95% CI: 86.4–94.2%), and specificity 91.7% (95% CI: 87.5–95.6%). CNNs, ResNet, and VGG-based architectures were the most frequently used (56% of studies). Applications spanned caries detection (mean AUC = 0.93), oral cancer identification (mean accuracy = 97.8%), periodontal bone loss analysis (AUC = 0.89), cephalometric landmark identification (AUC = 0.96), and implant planning (precision = 0.94).Meta-analytic synthesis revealed that AI models outperformed human experts in 68% of comparative studies, with statistically significant improvements (p < 0.05) in diagnostic sensitivity for image-based tasks. Conclusion: AI and ML demonstrate high diagnostic validity and reliability in dentistry, showing strong evidence of outperforming traditional assessment in early disease detection, classification, and treatment planning. However, standardized validation protocols, multi-center datasets, and explainable algorithms are crucial for clinical translation. AI should be regarded as an adjunct to—rather than a replacement for—clinical expertise, enabling precision-based, patient-centered dental care.

Artificial Intelligence and Machine Learning in Dentistry: Evidence-Based Perspectives — A Systematic Review

Key Points

Abstract

Cite This Study