Clinician-accessible automated machine learning in oral healthcare: A systematic review

Automated machine learning (AutoML) has emerged as a clinician-accessible approach to artificial intelligence by automating key stages of model development, including algorithm selection and hyperparameter optimization. This systematic review aimed to evaluate the application, performance, and translational relevance of AutoML in oral healthcare. Electronic searches of PubMed, Web of Science, Cochrane Library, and grey literature were conducted. Nineteen primary studies met the inclusion criteria. AutoML was applied across multiple dental specialties, including periodontology, orthodontics, pediatric dentistry, oral surgery, oral radiology, and public oral health, using heterogeneous data modalities such as radiographic images, clinical records, photographs, and omics data. Most studies reported strong performance on internal validation (e.g., cross-validation or validation splits) and/or independent hold-out test sets, particularly for imaging-based classification tasks, with several models achieving high accuracy and discrimination. However, external validation was uncommon, sample sizes were frequently limited, and substantial risk of bias was identified, particularly in analysis and participant selection. No study evaluated real-world clinical implementation or patient outcomes. Overall, AutoML shows promise in reducing technical barriers and supporting clinician-led artificial intelligence research in dentistry. Nevertheless, current evidence base remains exploratory, and rigorous external validation, standardized reporting, and prospective clinical evaluation are required before routine clinical adoption.