Date created: | Last Updated:

Category: Analysis

Description: Introduction: Artificial intelligence (AI) in dentistry can improve the diagnosis and prediction of oral diseases such as caries, periodontitis, endodontics, and oral cancer. The effectiveness of AI models depends on the quality of data and metadata, which describe data attributes, configurations, and contexts critical to the reproducibility and validity of research results. However, inconsistencies in metadata reporting hinder the development of robust AI models. Only 1.5% of dental articles share data, while 32.6% adhere to FAIR principles, highlighting the need for standardized metadata practices. Comprehensive metadata reporting ensures that AI models perform equitably across different populations and settings. Objective: This scoping review aims to identify current metadata reported in artificial intelligence-based dental research, identify gaps, characterize current reporting practices, and inform the development of METADENT, a reporting guideline for dental research metadata. Methods: A comprehensive search of three electronic databases (PubMed, IEEE Xplore, and ArXiv) will be conducted to identify published studies on AI applications in dentistry using a specific search strategy. Independent reviewers will screen the titles and abstracts of the collected studies against predetermined inclusion criteria. Studies that meet the criteria will undergo a full-text review before final selection. Data from selected studies will be extracted and duplicated by a team of researchers. Disagreements will be resolved by consensus with a third researcher. We will analyze the reported metadata using descriptive statistics, gap analysis, and comparative analysis. Results will be presented in tables and graphs. A narrative synthesis integrating quantitative and qualitative findings will be presented, and implications for future research and standardized metadata reporting guidelines will be discussed. Keywords: Metadata, Artificial Intelligence, Deep-learning models, Datasets.