Artificial Intelligence for Automated Grading and Treatment Planning in Periodontitis

Periodontitis is a widespread chronic inflammatory disease that continues to threaten oral health and contributes to systemic complications. Its diagnosis and grading largely depend on probing and radiographic assessment, yet these approaches vary across clinicians and lack precision in detecting early bone alterations. Deep learning has been introduced to address these shortcomings by automatically analysing dental images and extracting both global bone patterns and site-specific features relevant to disease severity. Encoder-decoder networks can delineate alveolar bone contours and periodontal pockets, while classification models combine these representations to generate reproducible grading outcomes. Compared with conventional methods, such systems offer more consistent evaluation of complex regions, reduce observer variability, and shorten the time required for clinical interpretation. Integration with structured reporting further facilitates incorporation into electronic health records, enabling routine use in follow-up and treatment planning. Remaining barriers include annotation inconsistency, equipment-related variability, and limited validation across centers. This review aims to synthesize current progress in deep learning-based grading of periodontitis, clarify unresolved challenges, and outline requirements for clinical adoption.