Application of artificial intelligence in the diagnosis of peri-implantitis and assessment of the prognosis of dental implantion. (A systematic review)

Dental implantology plays a central role in modern dentistry, but the long-term success of implant therapy largely depends on the health of peri-implant tissues. Peri-implantitis remains one of the most common causes of implant failure, characterized by soft tissue inflammation and progressive bone resorption. Conventional diagnostic methods, such as clinical examination and radiographic evaluation, are limited by low sensitivity at early stages and high operator dependence, resulting in reduced reproducibility and diagnostic accuracy. In recent years, artificial intelligence (AI) technologies, including machine learning, deep learning, and convolutional neural networks (CNN), have been increasingly applied in the diagnosis and prognosis of peri-implant diseases. Literature analysis demonstrated that AI-based models can detect subtle signs of bone resorption with accuracy comparable to or exceeding that of clinicians, provide standardized diagnostic protocols, and generate personalized treatment outcome predictions. Furthermore, integrating radiological findings with microbiological and immunological data (such as IL-1β and IL-6 levels) enhances risk stratification and allows the identification of patients with higher susceptibility to peri-implant disease progression. Nevertheless, several challenges hinder the routine clinical implementation of AI. Most studies are retrospective, involve small and heterogeneous samples, and use different architectures and performance metrics, limiting comparability and generalizability. Large-scale, multicenter, prospective studies with standardized evaluation protocols are required to validate and establish AI as a reliable tool in clinical implant dentistry.