Implementation of a predictive strategy in the diagnosis of inflammatory periodontal diseases

Relevance. The diagnosis of periodontal diseases, considering their severity, prevalence, progression, and staging, can be achieved by determining the levels of biomarkers or molecular imaging biomarkers in biofluids such as crevicular or sulcular fluid (GCF or GSF), saliva, and oral fluid. GCF is currently regarded as one of the diagnostically significant biological fluids for assessing the condition of periodontal tissues, not only in clinical diagnostic laboratories but also in dental offices. The implementation of sensitive, highly accurate, non-invasive, and specific methods for rapid GCF diagnosis, based on the qualitative analysis of biomarkers of cytokine imbalance, immunological disorders, changes in non-specific defence factors, and biophysical indicators, will allow for an objective assessment of the condition of periodontal tissues. Purpose. To improve the efficiency of periodontitis prevention using a developed mathematical model for personalized prediction of the course of inflammatory periodontal diseases based on the investigated biomarkers in GCF. Material and methods. The study included 101 patients: Group I consisted of 22 patients diagnosed with K05.10 (gingivitis), Group II included 31 patients diagnosed with K05.31 (mild periodontitis), and Group III comprised 18 patients diagnosed with K05.31 (moderate periodontitis). The comparison group consisted of 30 individuals with clinically healthy periodontium. All subjects underwent clinical and instrumental examination, determination of periodontal indices, GCF collection, and quantitative analysis of immune regulatory mediators (IL-1β, IL-6, TNF-α, IL-8, MCP-1, IL-17, VEGF, IL-1RA). Results. The study of immune regulatory mediators confirmed the significance of increased levels of pro- and anti-inflammatory cytokines/chemokines, as well as the reduction of the anti-inflammatory biomarker IL-1RA in GCF at the early stages of inflammatory changes in periodontal tissues. This is accompanied by the appearance of signs indicating the destruction of the dentogingival junction. Using logistic regression and training a multiclass classifier based on the support vector machine method, a model was developed to predict the risk of dentogingival junction loss in patients, potentially leading to periodontitis. Conclusion. The results of logistic regression modelling and training a multiclass classifier based on the support vector machine method demonstrate that in diagnosing the initial stages of periodontal tissue damage with the loss of the dentogingival junction (DGJ), the most effective approach is the comprehensive use of inflammatory process biomarkers and the development of multi-marker algorithms based on a computer program.