Comparison of diagnoses made by dentistry students and by artificial intelligence dentists

Authors:
Publisher website
Google Scholar
Abstract
Objectives: The introduction of artificial intelligence (AI) about great changes in the field of dentistry, but it has not yet been fully determined in which areas it will make a positive contribution to dentistry students. The objective of our study was to compare the diagnostic accuracy of undergraduate students (fourth‐year dentistry students [4DS] and final‐year dentistry students [5DS]) and AI when examining panoramic radiographs.Methods: Fifty panoramic radiographs and 1602 teeth were examined by 50 4DS who had not received a clinical practice internship, 50 5DS, and an AI application. The participants and the AI application evaluated the teeth seen in each radiograph one by one in terms of caries, fillings, teeth with root canal treatment, periodontal loss, extractions, crowns, teeth with apical lesions, and impacted and extracted teeth. Findings were recorded in an Excel chart. Chi‐square analysis was used to compare diagnostic success between the groups.Results: The results indicate that there was a statistically significant difference in the identified accuracy of caries, fillings, and extractions between the AI application and undergraduate students (p < 0.05). Although AI showed more identified accuracy in teeth with apical lesions, impacted teeth, and teeth with root canal treatment than in undergraduate students, there was no significant difference between them (p > 0.05).Conclusion: AI exhibited better results than undergraduate students especially in the detection of caries and fillings. AI could improve undergraduates’ accuracy in detecting caries, fillings, and extractions and help them make accurate treatment decisions. In cases where dentistry students are examining patients using panoramic radiographs, employing AI programs during their clinical training to confirm and strengthen the student's diagnosis may be a promising new development.

This publication has 30 references indexed in Scilit: