Abstract

This systematic review assessed the efficacy of artificial intelligence (AI) models in predicting the progression of periodontal disease, a critical capability for enabling timely intervention and personalized treatment plans in dental care. A systematic literature search was performed across multiple electronic databases to identify relevant studies. The review protocol was guided by a focused search question on AI’s role in prognostic prediction using clinical and radiographic data. Study selection, data extraction, and quality assessment were conducted independently. The review included nine studies with diverse designs, from small clinical trials to large-scale retrospective analyses. AI models, such as Random Forest and complex architectures like probabilistic graphical models and convolutional neural networks, demonstrated superior performance over traditional statistical methods. These models achieved high predictive accuracy, with area under the curve values ranging from 0.74 to 0.99, by integrating multimodal data. Key predictors of disease progression and tooth loss included probing pocket depth, clinical attachment level, bone loss, and specific biomarkers. Studies also showcased the successful application of natural language processing for automated disease tracking from electronic dental records with near-perfect accuracy (98%-99%). However, the review identified a “black box” problem, with many high-performing models lacking interpretability, and a moderate risk of bias in most studies due to incomplete external validation and heterogeneous methodologies. AI showed strong potential for revolutionizing the prediction of periodontal disease progression. Future research must prioritize clinical validation, standardized outcome definitions, enhanced model interpretability, and focus on seamless integration into clinical workflows.

Key words: Artificial intelligence, periodontal diseases, machine learning, disease progression, deep learning, natural language processing, systematic review