ARTIFICIAL INTELLIGENCE IN EARLY DETECTION OF TEMPOROMANDIBULAR JOINT DISORDERS-A SYSTEMATIC REVIEW

Background: Temporomandibular joint disorders (TMDs) affect a significant portion of the population and are a leading cause of chronic orofacial pain and functional limitation. Early diagnosis is crucial for effective intervention, yet conventional diagnostic methods often fall short in accuracy and accessibility. Recent advancements in artificial intelligence (AI) offer a novel approach to early detection through enhanced image analysis, but existing evidence is scattered and lacks systematic synthesis. Objective: This systematic review aims to evaluate the effectiveness and diagnostic performance of AI-based tools in the early identification of temporomandibular joint disorders. Methods: A systematic review was conducted following PRISMA guidelines. Databases searched included PubMed, Scopus, Web of Science, and the Cochrane Library, covering studies published between January 2018 and April 2024. Inclusion criteria encompassed human studies utilizing AI for TMD diagnosis through imaging modalities such as MRI, CBCT, or panoramic radiographs. Exclusion criteria included non-English articles, animal studies, and reviews. Data extraction focused on study design, population, AI model used, imaging type, and diagnostic outcomes. Risk of bias was assessed using the Newcastle-Ottawa Scale and Cochrane tools. Results: Eight studies involving 2,138 participants were included. AI models—primarily convolutional neural networks and deep learning systems—achieved high diagnostic performance with accuracy ranging from 85.7% to 92.3%, sensitivity between 88.0% and 94.1%, and AUC values up to 0.96. Most tools matched or exceeded the diagnostic capabilities of human experts. Risk of bias was low to moderate, though some concerns regarding model validation and blinding were noted. Conclusion: AI-based diagnostic systems demonstrate strong potential for early and accurate detection of TMDs, offering a valuable adjunct to clinical decision-making. However, larger, externally validated studies are needed to support widespread clinical implementation and ensure reproducibility.