The Impact of Artificial Intelligence on Contemporary Orthodontic Treatment Planning - A Systematic Review and Meta-Analysis
M’hamed Jihed, Inès Dallel, Samir Tobji, Adel Ben Amor
Abstract
Objective: To assess available evidence on the use of artificial intelligence (AI) in the planning of customized orthodontic therapy. The aim of the meta-analysis was to evaluate the performance and effectiveness of AI models for orthodontic treatment planning and decision-making. Materials and methods: PubMed, EBSCO host, ScienceDirect, Scopus, and Web of Science were searched over the period from January 1, 2000 to January 9, 2021, then they were updated until January 19, 2022. A systematic review and diagnostic test accuracy meta-analysis were performed. Results: Overall, 1037 records were identified. A total of twelve studies were ultimately included in the qualitative synthesis, of which five studies were included in the meta-analysis. Pooled sensitivity, specificity, diagnostic odds ratio, and area under the curve with 95% confidence intervals of AI models’ performance were: 0.965 (0.921-0.985), 0.962 (0878-0.989), 695.537 (232.742-2078.572), 0.99 (0.98-1.00), respectively. The accuracy of AI systems reached 95.47%. Conclusions: The findings show promising results concerning the diagnostic accuracy of AI systems for orthodontic treatment planning and decision-making and its implementation in clinical settings. AI models are successful in predicting valid treatment plans with accurate decisions. Thus, they can ease global treatment and improve outcomes.