MULTIFACTOR REGRESSION MODEL FOR PREDICTION OF CHRONIC RHINOSINUSITIS RECURRENCE

The aim: To propose an approach to forecasting the risk of chronic rhinosinusitis recurrence based on multivariate regression analysis for effective diagnosis and carrying out treatment and preventive measures.

Materials and methods: 104 patients aged 18 to 80, including 58 women and 46 men, diagnosed with chronic rhinosinusitis were examined.

Results: To build a multifactorial regression model for predicting the recurrence of chronic rhinosinusitis, probable factors of the occurrence of the disease were selected. 14 possible factors were analyzed using multivariate regression analysis. 13 risk factors were selected for predicting recurrence of chronic rhinosinusitis with a significance level of less than 0.05. Histograms of the residual deviations of predicting the recurrence of chronic rhinosinusitis were obtained, which are distributed symmetrically, and a normal-probability straight line is presented, on which there are no systematic deviations. The given results confirm the statistical hypothesis that the residual deviations correspond to the normal distribution law. Residual deviations relative to the predicted values are scattered chaotically, which indicates the absence of dependence on the predicted values of the risk of recurrence of chronic rhinosinusitis. The value of the coefficient of determination was calculated, which is 0.988, which gives grounds to claim that 98.8% of the factors are taken into account in the model for predicting the recurrence of chronic rhinosinusitis and its high reliability and acceptability in general.

Conclusions: The proposed model makes it possible to predict in advance potential complications and the possibility of recurrence of the studied disease.