AIによる歯科モニタリングの一般的な歯科矯正治療関連の緊急事態検出の精度

This study aimed to evaluate the diagnostic performance of DentalMonitoring™ (DM) (Dental Monitoring SAS, Paris, France), an FDA-cleared AI-powered orthodontic remote monitoring software, in detecting three common orthodontic appliance issues: bracket debonding, open self-ligating clips, and tie loss. Datasets from 1,014 US-based patients were analyzed. Each DM image set was assessed by the AI algorithm and independently reviewed by a panel of three orthodontic experts, whose consensus served as the reference standard. A total of 659 evaluations were included for bracket debonding, 647 for self-ligating clip status, and 653 for tie presence. Sensitivity and specificity along with their 95% confidence intervals were calculated using a two-level evaluation basis (positive vs. negative) across the three clinical parameters. DM's AI demonstrated high diagnostic performance, with sensitivity of 98.4% for bracket debonding, 91.1% for open self-ligating clips, and 93.3% for tie loss. Corresponding specificity values were 99.6%, 88.3%, and 96.5%, respectively. Current results indicate that DM's AI analysis system has high accuracy in detecting bracket debonding, open self-ligating clips, and tie loss. DM can help significantly reduce the rate of these undetected clinical incidents, providing a better approach to managing emergencies and maintaining clinical control in orthodontic treatment.