Tipus de document
ArticleVersió
Versió publicadaData de publicació
Llicència de publicació
Si us plau utilitzeu sempre aquest identificador per citar o enllaçar aquest document: https://hdl.handle.net/2445/230251
Precisión diagnóstica de la inteligencia artificial en el diagnóstico de caries: revisión sistemática
Títol de la revista
Director/Tutor
Contribució addicional
ISSN de la revista
Títol del volum
Resum
Background: Dental caries is among the most prevalent oral diseases worldwide. In recent years, artificial intelligence (Al), particularly convolutional neural networks (CNNs), has been proposed as a supportive tool to improve diagnostic accuracy for caries detection from diagnostic images. Methods: A systematic review with qualitative synthesis was conducted in PubMed, including studies published between 2020 and 2025. The last search was performed on May 30, 2025. Human studies with full-text access applying AI to caries diagnosis and reporting diagnostic accuracy metrics were included. Methodological quality was assessed using AMSTAR-2 for systematic reviews and QUADAS-2 for primary diagnostic accuracy studies. Results: Ten studies were included. Reported sensitivities ranged from 25% to 99.8% (predominantly similar to 44%-94%) and specificities from 16.46% to 100% (mostly similar to 77%-98%). Five studies reported AUC values between 0.74 and 0.99. Bitewing radiographs were the most commonly used modality. Main architectures included U-Net, ResNet, and YOLO. Conclusion: AI is a promising adjunct tool for caries diagnosis. Nevertheless, current evidence shows methodological heterogeneity and limited external validation; therefore, standardized protocols, multicenter validation, and professional training are required for safe clinical implementation.
Recurs relacionat
Matèries (anglès)
Citació
Citació
NUNHA, Selina, et al. Precisión diagnóstica de la inteligencia artificial en el diagnóstico de caries: revisión sistemática. Academic Journal of Health Sciences. 2026. Vol. 41, num. 2, pags. 267-278. [consulted: 21 of November of 2026]. Available at: https://hdl.handle.net/2445/230251