Artificial intelligence in periodontics and peri-implant medicine: from diagnosis to precision care

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Background: Artificial intelligence (AI) has substantial potential to advance periodontal and peri- implant care through novel diagnostic, prognostic, and therapeutic opportunities. Nevertheless, the evidence is scattered over several fields. Objective: To provide a narrative synthesis of the recent evidence regarding applications of AI to periodontology and peri-implant medicine specifically diagnostic imaging, disease prediction, biomarker discovery, peri-implant disease assessment and clinical decision support. Methods: A systematic search of PubMed, MEDLINE, Scopus, and Embase for studies from August 2020 to August 2025 was performed. We included only original research articles on AI applications in periodontal or peri-implant settings. After screening the identified records and conducting full-text assessments, a total of 32 studies were included based on established inclusion criteria. Results: AI showed high diagnostic performance at imaging-based applications, with accuracy from 0.85 to 0.98 reported for the detection and quantification of alveolar bone loss using two- and three-dimensional radiographs. Using this data, machine learning (ML) models were developed for identifying potential molecular biomarkers that could support disease prediction and gain insight into biological processes. AI-aided immune profiling and risk stratification models demonstrated robust predictive capabilities in peri-implant maintenance. mHealth is one of the types of AI-based digital health devices and has succeeded in enhancing patient compliance and periodontal prognosis. Conclusions: AI has considerable promise to impact non-invasive periodontal and peri-implant diagnostics, risk-assessment and enhanced treatment-planning through advanced imaging analysis, molecular profiling as well as patient-centred digital tools. These need to be further validated, standardized, and integrated into clinical care before widespread uptake.

This publication has 38 references indexed in Scilit: