Scholarlink

Aim/Objective <p>This review assessed the quality and findings of systematic reviews on AI in dentistry, categorising advancements across various specialties.</p> Methods <p>The review analyzed data from seven databases, assessed review quality with ROBIS, calculated pooled AI performance estimates, and identified research gaps through an Evidence Gap Map.</p> Results <p>This study analysed 116 included systematic reviews. Meta-analysis of twelve low-bias reviews showed AI diagnostic accuracy ranging from 82% to 95% across dental specialties. The pooled sensitivity and specificity of AI algorithms for dental diagnostics were 0.85 (95% CI: 0.76–0.93) and 0.93 (95% CI: 0.90–0.95), respectively. Advanced models, particularly Convolutional Neural Networks (CNN), demonstrated a pooled accuracy of 93.1% (95% CI: 91.19–95.05%). Corrected Covered Area analysis indicated low overlap among reviews (10%), reflecting the diverse applications of AI in dentistry. Significant heterogeneity across pooled sensitivity (I<sup>2</sup> = 98.26%), specificity (I<sup>2</sup> = 87.49%), area under the curve (I<sup>2</sup> = 86.62%) and accuracy (I<sup>2</sup> = 75.86%) were observed.</p> Discussion <p>AI shows strong diagnostic accuracy across dental specialties like caries detection, cephalometric landmark identification, and oral lesion diagnosis, with pooled sensitivity (0.85), specificity (0.93), and AUC (0.95) values. Advanced AI models like CNNs, Artificial Neural Networks, and larger, diverse datasets improve diagnostic accuracy, especially in image classification. Addressing research gaps and standardising methods are key to optimizing AI’s clinical impact.</p> Conclusion <p>This review reinforces AI’s transformative potential in dentistry, enhancing tasks like diagnosis, detection, and prognosis, particularly in caries and lesion detection, to improve clinical decision-making and patient outcomes.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Artificial Intelligence in dentistry: an overview of systematic reviews and meta-analysis

  • 期刊论文

  • 发表日期：

    2025年12月9日

查看全文
  • Ankita Saikia,
  • Therese Kvist,
  • Amr Fawzy,
  • Robert Anthonappa

摘要

Aim/Objective

This review assessed the quality and findings of systematic reviews on AI in dentistry, categorising advancements across various specialties.

Methods

The review analyzed data from seven databases, assessed review quality with ROBIS, calculated pooled AI performance estimates, and identified research gaps through an Evidence Gap Map.

Results

This study analysed 116 included systematic reviews. Meta-analysis of twelve low-bias reviews showed AI diagnostic accuracy ranging from 82% to 95% across dental specialties. The pooled sensitivity and specificity of AI algorithms for dental diagnostics were 0.85 (95% CI: 0.76–0.93) and 0.93 (95% CI: 0.90–0.95), respectively. Advanced models, particularly Convolutional Neural Networks (CNN), demonstrated a pooled accuracy of 93.1% (95% CI: 91.19–95.05%). Corrected Covered Area analysis indicated low overlap among reviews (10%), reflecting the diverse applications of AI in dentistry. Significant heterogeneity across pooled sensitivity (I2 = 98.26%), specificity (I2 = 87.49%), area under the curve (I2 = 86.62%) and accuracy (I2 = 75.86%) were observed.

Discussion

AI shows strong diagnostic accuracy across dental specialties like caries detection, cephalometric landmark identification, and oral lesion diagnosis, with pooled sensitivity (0.85), specificity (0.93), and AUC (0.95) values. Advanced AI models like CNNs, Artificial Neural Networks, and larger, diverse datasets improve diagnostic accuracy, especially in image classification. Addressing research gaps and standardising methods are key to optimizing AI’s clinical impact.

Conclusion

This review reinforces AI’s transformative potential in dentistry, enhancing tasks like diagnosis, detection, and prognosis, particularly in caries and lesion detection, to improve clinical decision-making and patient outcomes.