Abstract
PURPOSE: Tooth-level prognostic systems can be used for treatment planning and help to identify teeth at risk of being lost over time. However, several sets of criteria were proposed to estimate tooth prognosis in the context of periodontitis patients. OBJECTIVES: To identify the prognosis tools available and to evaluate their relevance in predicting periodontal-related tooth loss (TLP). METHODS AND MATERIALS: An electronic search was conducted for published data in MEDLINE, EMBASE, COCHRANE up to January 2025. Reference lists of retrieved studies for full-text screening and reviews were hand-searched for potentially eligible studies. RESULTS: In total, 1,471 records were identified from databases, and an additional 7 unique records were identified through citation searching of selected studies from the database search. After screening, 33 studies were selected for full-text review, of which 22 were included in the final selection, with 6 classical models, 11 regression-based models, 2 AI-driven prognostic models and 3 external validations. Most prognostic models confirm high precision and accuracy, fluctuating in most cases around a value of area under the curve (AUC) = 0.8. CONCLUSIONS: Patient (smoking, diabetes) and tooth-related factors (furcation involvement, increased probing depth, mobility) influence the prognosis of tooth retention in the long term. The identification of prognostic factors is of crucial importance to better predict the long-term survival of teeth and to adapt the treatment plan. All prognostic systems at the dental level showed excellent predictive capacity for the risk of tooth loss linked to periodontitis. An AI-based machine-learning algorithm will be a helpful tool for determining tooth prognosis.