AI-driven digital workflow for personalized orthodontic aligners with real-time biomechanical feedback

Thesis Abstract

This study addresses the persistent challenge of achieving efficient, patient-specific orthodontic tooth movement while minimizing treatment time and adverse effects through an AI-enhanced digital workflow that integrates personalized aligner design with real-time biomechanical feedback. The aim is to develop and validate a closed-loop system that combines machine learning–driven prediction of aligner forces with sensor-augmented aligners to continuously monitor tooth displacement, force delivery, and tissue response. Specific objectives include (i) to develop a virtual patient model that personalizes material properties, bracket- and aligner-tooth interactions, and tooth movement trajectories using finite element analysis and reinforcement learning; (ii) to implement a real-time sensor suite within clear aligners capable of measuring intraoral forces, fit accuracy, and occlusal contacts, and to correlate these data with predicted biomechanical outcomes; (iii) to evaluate the predictive performance of the AI workflow against conventional planning in a pilot randomized controlled trial; and (iv) to assess treatment efficiency, patient comfort, and clinician workload across diverse malocclusion severities. The methodological approach adopts a mixed-methods, multi-center design. The population comprises orthodontic patients with mild to moderate malocclusions (N = 120) recruited from three tertiary dental clinics, randomized into AI-assisted and standard workflow groups. Data collection employs triangulated instruments digital intraoral scans, 3D surface tracking, and embedded micro-sensors within aligners to capture force vectors, contact pressures, and wear. Additional data include clinical treatment progress records, patient-reported outcome measures (PROMs), and clinician assessments. Validity and reliability of instruments are ensured through calibration protocols, repeatability studies, and cross-validation of sensor outputs against finite element simulations. Data analysis methods encompass machine learning regression and time-series analyses to predict tooth displacement from sensor-derived biomechanical features; multivariate regression and mixed-effects models to compare treatment outcomes between groups; ANOVA to examine differences across malocclusion subtypes; and thematic analysis of clinician and patient interviews to illuminate usability and perceived value. A conceptual framework grounded in Biomechanical Theory of Tooth Movement and the Technology Acceptance Model informs model specification, with a secondary basis in the Theory of Planned Behavior to interpret stakeholder adoption patterns. The study anticipates that the AI-driven workflow will yield higher prediction accuracy of tooth movement trajectories (expected R2 > 0.85 in hold-out tests), reduced average treatment duration by 12–18%, and improved aligner fit satisfaction scores by at least 20% relative to conventional planning. Additional expected findings include enhanced control of overcorrection and reduced unwanted tooth movement due to real-time feedback loops, along with increased clinician confidence in treatment sequencing. The theoretical contribution lies in operationalizing an integrated AI–Biomechanics framework that merges predictive analytics with real-time sensor data to optimize orthodontic force delivery, thereby extending the application of reinforcement learning in clinical biomechanics. Practically, the study offers a scalable, digital workflow enabling personalized aligner production with closed-loop monitoring, potentially standardizing treatment quality and enabling remote oversight. The limitations anticipated involve sensor durability within the oral environment, variability in patient compliance, and the generalizability of findings to complex malocclusions beyond mild-to-moderate cases. Policy and practice implications include implications for regulatory approval of smart dental devices, data governance for intraoral sensing, and the need for updated clinical guidelines that incorporate AI-assisted treatment planning and monitoring. The study concludes that an AI-driven, sensor-enabled digital workflow can significantly improve the precision and efficiency of clear aligner therapy, while enhancing patient experience and clinician decision support; recommendations emphasize iterative refinement of sensor technology, larger-scale multicenter trials across broader malocclusion spectra, and integration with teleorthodontics platforms to broaden access and continuous care.