ARTIFICIAL INTELLIGENCE IN ORTHODONTICS: DIAGNOSIS, DISPUTE RESOLUTION, AND OUTCOME PREDICTION

Résumé fourni par la source

Artificial intelligence (AI) is increasingly influencing orthodontic diagnosis, treatment planning, clinical communication, and prediction of treatment outcomes. Machine-learning and deep-learning systems can analyze radiographs, photographs, three-dimensional scans, and clinical records to identify anatomical landmarks, classify malocclusions, detect impacted teeth, estimate skeletal relationships, and forecast changes associated with orthodontic treatment. This article reviews the role of AI in orthodontics with particular attention to three applications: diagnosis, dispute resolution, and outcome prediction. A narrative review was conducted using recent scientific literature, systematic reviews, clinical studies, and international guidance on AI governance in healthcare. The available evidence suggests that AI systems can achieve moderate to high diagnostic performance, although reported accuracy varies substantially according to the task, dataset, image quality, reference standard, and degree of external validation. In orthodontic diagnosis, convolutional neural networks have demonstrated promising performance in cephalometric landmark identification, skeletal classification, extraction-need assessment, and radiographic interpretation. In dispute resolution, AI may support the reconstruction of treatment decisions, comparison of pre-treatment and post-treatment records, identification of deviations from an agreed plan, and transparent documentation of communication. However, AI should function as an evidentiary and decision-support instrument rather than an autonomous legal or clinical judge. Resonsible implementation requires human oversight, informed consent, secure data governance, model validation, and continuous auditing. AI is likely to strengthen orthodontic care when integrated with professional expertise, but it cannot replace clinical judgment or the patient–orthodontist relationship.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.