AI-enhanced 3D tooth movement forecasting in clear aligner therapy using deep morphometric modelling: A prospective validation study

Abstract

Clear aligner therapy often encounters early tracking deviations that compromise treatment efficiency, creating a need for predictive tools that identify risk at the outset. Therefore, it is of interest to develop and validate a deep morphometric AI model capable of forecasting early aligner tracking deviation using baseline and first-week 3D intraoral scans. Hence, a prospective sample of 40 adults was analyzed using a graph-convolutional neural network trained on geometric mesh features extracted from sequential scans. The model demonstrated strong performance, achieving 85% accuracy with an RMSE of 0.19 mm in predicting clinically significant early deviation. Thus, we show that AI-driven morphometric analysis offers a promising approach for early risk detection and improved treatment planning in clear aligner therapy.