A multifactorial model for early prediction of chronic periodontitis risk in patients with metabolic-associated steatotic liver disease

We are aware of the important pathogenetic link between oral cavity health and systemic metabolic processes, particularly between chronic periodontitis and steatotic liver disease, in which the metabolic component acts as a modifier of the inflammatory response. The identification of predictors and the development of predictive models will help to identify the risks of development at an early preclinical stage. Objective — to develop an effective mathematical model for predicting the risk of early periodontal inflammation and create an algorithm for screening stratification of patients with metabolic‑associated steatotic liver disease. Materials and methods. a diagnostic study of 259 clinical cases was carried out, and after auditing the data, 150 valid episodes were included in the statistical analysis. A randomization method was applied to split the sample into a training set for developing the regression equation (n=120) and a test set to validate the accuracy of the resulting algorithm (n=30), using an 80 : 20 ratio. Model construction was performed using stepwise logistic regression, with evaluation based on the Wald test, Nagelkerke’s R‑squared, the Hosmer‑Lemeshow test, and ROC analysis (AUC). Risk visualization is presented using a radar chart. Results. Using the stepwise inclusion method for mathematical modeling, a regression function was constructed, and the five most significant predictors were identified: threshold densitometry, salivary viscosity, presence of MASLD, gingival crevicular blood glucose level, and body mass index (BMI). In the resulting regression equation, the Nagelkerke R‑squared was 0.887 (88.7% of the variance), and the statistical significance of the model was confirmed by the log‑likelihood of –2 (G=29.463; p=0.0013). The model demonstrated a sensitivity of 100% and a specificity of 91.2%. For clinical application, a radar chart was proposed with a cut‑off point of 3, where exceeding this threshold indicates a high risk of developing periodontal lesions. Conclusions. The proposed predictive model serves as a reliable tool for early personalized screening of periodontal lesions in patients with metabolically associated steatotic liver disease.