Microbiota as a prognostic predictor of the risk of peri-implantitis development in patients with chronic periodontitis

Development of inflammation at the site of the installed implant, with the development of peri-implantitis and mucositis are much more common in patients with chronic periodontal diseases. The study is aimed at developing a model for predicting the risk of peri-implantitis in patients with chronic periodontitis using microbiological markers and the CART algorithm. Objective: to develop a model for classifying the risk of peri-implantitis in patients with chronic periodontitis based on microbiological markers using the decision tree method. Materials and methods. The study analyzed 177 patients with chronic periodontitis, divided into three groups: without dental implants, and implants without signs of peri-implantitis and with the presence of these signs (59 people in each group). An analysis of microbiological indicators was carried out, which allowed us to identify microbiological markers of the risk of peri-implantitis. A machine learning model was built based on the CART algorithm with the inclusion of the Gini index. Results. For the total sample, the model demonstrated the following metrics: overall correctness — 0.667, consistency — 0.639, ROC—AUC — 0.66. Statistically significant differences in microbiological parameters were revealed in patients with peri-implantitis. The most typical microorganisms were Rothia mucilaginosa, Actinomyces odontolyticus, Staphylococcus epidermidis, Streptococcus australis, Streptococcus oralis. These results are consistent with the literature data. Conclusions. Microbiological analysis of patients with peri-implantitis, including statistical data processing and the use of machine learning methods, allows us to predict the risk of peri-implantitis in patients with chronic periodontitis. The results emphasize the importance of including repeat microbiological examination in standard examination protocols, which allow for early identification of at-risk patients and personalization of preventive and therapeutic measures.