Beyond Model Performance: From Algorithmic Innovation to Clinical Translation of Artificial Intelligence in Oral Healthcare

Artificial intelligence (AI) is advancing dental imaging diagnosis, risk assessment, treatment planning, and oral health promotion. Multimodal learning, transformers, foundation models, federated learning, and agentic systems have been used in many clinically relevant tasks in oral healthcare. However, most dental AI research remains retrospective, single-centre, and focused on internal test-set performance. High accuracy alone does not establish clinical effectiveness, clinical utility, patient benefit, and translational value of AI tools in real-world settings. Translating dental AI algorithms into real-world use requires a continuous evidence chain connecting clinically grounded data and algorithm development with multicentre validation, prospective evaluation, workflow integration, and post-deployment oversight.However, high-quality multicentre datasets are scarce, and relatively few studies report performance on external, independent test sets. This limited evidence leaves clinicians uncertain about the real-world clinical effectiveness of these AI systems, particularly their effects on patient-reported and clinician-oriented outcomes. Robust data from routine clinical use also remain limited. These evidence gaps hinder the clinical translation and adoption of innovative AI algorithms and software across oral disease prevention, diagnosis, and treatment.The goal is to build a shared evidence framework connecting dental AI development with translational clinical and societal value. This Research Topic welcomes research across the full translational spectrum of AI in oral healthcare, from novel algorithms, datasets, and multimodal foundation models to independent and multicentre validation, prospective clinical evaluation, workflow implementation, comparative effectiveness, and post-deployment monitoring.• Algorithm and dataset development: including clinically grounded dental tasks, representative datasets, multimodal and foundation models, federated learning, and robust reference standards.• External and multicenter validation: examining the real-world value of AI systems and software.• Prospective clinical evaluation: of human-AI performance, downstream decisions, patient-reported and clinician-reported outcomes, and unintended overdiagnosis or overtreatment.• Clinical trials examining dental AI studies: clinical effectiveness research of AI in dentistry, various clinical trials (including retrospective, cohort study, prospective studies and RCT).• Patient-reported or clinician-reported outcome: for AI related systems and tools.• Scoping review or systematic review of AI development, validation, prospective evaluation, clinical effectiveness, and implementation in dentistry. Topic editors: Zhuohong Gong, Dawei Xiang, Mengru Shi, Hengyi LIU Source: https://www.frontiersin.org/research-topics/86175/beyond-model-performance-from-algorithmic-innovation-to-clinical-translation-of-artificial-intelligence-in-oral-healthcare