Background
Artificial intelligence is largely incorporated into dental research and clinical work. Predicting tooth restorability using artificial intelligence would help inexperienced dental professionals in clinical decision-making and education. This study evaluates the agreement in predicting tooth restorability using parallel periapical radiographs between dental interns and a newly developed artificial intelligence model created and trained to predict tooth restorability.
Materials and Methods
A deep learning model was created and trained using a dataset of 53,035 radiographs according to a tooth restorability prediction score from parallel periapical radiographs developed by six dental specialists. Seventeen radiographs were chosen randomly, and the specialists assessed the restorability of each tooth twice, two weeks apart, per the scoring index they developed. The intra-rater and inter-rater reliability of the specialists were assessed using Cohen's kappa correlation coefficients. The radiographs were then independently assessed once by artificial intelligence and by 20 dental interns. The inter-rater reliability of the responses of the two groups was assessed using Cohen's and Fleiss's kappa correlation coefficients using SPSS and Excel for statistical analysis (p < 0.05).
Results
A perfect level of agreement was found between and among the specialists' responses, with the coefficient of kappa calculated to be 1. However, there was slight disagreement among the dental interns, with the Fleiss kappa calculated to be -0.03832. There was a statistically significant strong agreement between the dental interns and artificial intelligence responses, as the kappa coefficient was calculated to be 0.883, with a p-value < 0.01.
Conclusion
The performance of the developed and trained preliminary artificial intelligence model was similar to or even better than that of dental interns in predicting tooth restorability using parallel periapical radiographs.
Trial Registration:
Research Centre of Prince Sattam bin Abdulaziz University approval number: SCBR-178/2023 on November 05, 2023, and the institutional review board of Riyadh Second Health Cluster approval number: FWA00018774 on April 22. 2024.