LOGISTIC REGRESSION MODEL AS AN ADDITIONAL MATHEMATICAL METHOD FOR PREDICTING THE DEVELOPMENT OF lONE ATRIAL FIBRILLATION

Abstract

After studying the Association of the TT genotype and the T allele with the development of lone atrial fibrillation (AF) and constructing a logistic regression model, it was revealed that the probability share of lone AF in the presence of the TT genotype is 6,2%; for the absence of lone AF – 95,1%. The model predicts the development of lone AF with a sensitivity of 17,7%. The overall percentage of correct predictions is 65,4%. Background. To study the prediction of the probability of occurrence of lone AF depending on the predictor genotype of polymorphism rs2200733 in chromosome 4q25 using the method of multiple logistic regression and to determine the measure of statistically significant influence of the predictor genotype on the development of pathology. Materials and methods. 247 patients with AF (113 with lone and 134 with secondary) were examined. The control group was represented by 182 healthy people. All of them were performed a specific range of functional and laboratory methods of research, including molecular genetics. Logistic regression analysis was used to predict the development of lone AF. Results. According to the odds ratio, the presence of the TT genotype increases the risk of AF by 1,8 times (in the presence of the T allele, the risk of lone AF increases by 2,0 times), a logistic regression model is constructed. Model determination coefficient R2=0,062. The specificity of the model in terms of predicting the absence of an lone form of AF is 95,1%. The model predicts the development of an lone form of AF with a sensitivity of 17,7%. The total percentage of correct predictions is 65,4%. Conclusion. According to the logistic regression model, the probability of lone AF in the presence of the patient’s TT genotype is 6,2%; the absence of lone AF – 95,1%. The total percentage of correct predictions is 65,4%.

Keywords

• изолированная фибрилляция предсердий
• логистическая регрессионная модель
• полиморфизм rs2200733