An Analysis of an AI Analytical Tool for Predicting Caries Progression from Serial Intraoral Scans in Orthodontic Patients

Abstract

Background: Orthodontic treatment increases the risk of enamel demineralization due to plaque accumulation around fixed appliances, while conventional monitoring methods often fail to detect early-stage changes. Artificial intelligence (AI) applied to serial intraoral scans offers potential for enhanced longitudinal assessment and early detection of caries progression. Objective: To evaluate whether AI-guided analysis of serial intraoral scans improves early preventive intervention rates and reduces enamel demineralization progression compared with standard clinical monitoring in orthodontic patients. Methods: A randomized controlled trial was conducted among 72 orthodontic patients aged 12–25 years undergoing fixed appliance therapy, with 68 completing six-month follow-up. Participants were allocated to AI-guided monitoring or standard care. Serial intraoral scans were obtained at baseline, three months, and six months. Primary outcome was the rate of early preventive interventions, while secondary outcomes included time to intervention, changes in ICDAS-adapted enamel scores, and lesion depth progression. Statistical analyses included t-tests, chi-square tests, and Pearson correlation. Results: Early preventive interventions were significantly higher in the AI group (70.6%) compared with controls (38.2%) (p = 0.008), with shorter time to intervention (3.1 ± 1.2 vs. 4.4 ± 1.5 months; p = 0.003). The AI group showed lower enamel demineralization progression (ICDAS change: 0.42 ± 0.31 vs. 0.78 ± 0.46; p = 0.001; lesion depth: 0.12 ± 0.09 mm vs. 0.26 ± 0.14 mm; p < 0.001). AI risk scores correlated with enamel changes (r = 0.64, p < 0.001). Conclusion: AI-guided monitoring of serial intraoral scans enhances early detection and supports timely preventive care, significantly reducing enamel demineralization progression in orthodontic patients.