استفاده از مدل رده‌بندی درختی برای تعیین عوامل مؤثر بر مرگ‌ومیر پس از عمل جراحی کرونری بای‌پاس در بیماران غیر وابسته به دیالیز

 Background and Objective: Coronary artery disease is one of the most prevalent causes of death. A coronary artery bypass surgery is a common treatment for this disease. In addition, renal dysfunction can lead to increased mortality and post-operative complications. This study aimed to identify the most important factors influencing the mortality of patients who suffer from coronary artery disease and to introduce a classification approach according to Classification Tree (CART) model for predicting the mortality from this disease.    Materials and Methods: This research was conducted based on the information gathered from a cross-sectional study on 1390 patients (except dialysis-dependent) who undergone coronary artery bypass grafting, admitted to Cardiology ward of Shariati hospital during the years 2007-2010. The ordinary logistic regression model and a classification tree were utilized for predicting the probability of death in these patients. The SPSS version 18.0 and CART version 6.0 were used for data analysis.  Results: In this study, the classification tree model (CART) resulted in an accuracy of 90%. The patients with renal insufficiency, intra-aortic balloon pump placement during and after surgery, prolonged ventilation, and perfusion time over 160 were shown as the high-risk groups, while those patients with heart ventricular post-operative complications regarded as the medium-risk group. The sensitivity and specificity indices for this model were 82% and 89%, respectively, while it was 80.4% and 88%, respectively, for logistic model.   Conclusion: In the present study, the logistic and decision tree models led to nearly similar results, however, the decision tree model seemed to be more accurate. The IABP (Intra-Aortic Balloon Pump) was the most effective factor for mortality. The mortality rate due to this factor during and after surgery for all patients was 19% and 54.1%, respectively.