One-year renal graft survival prediction using a weighted decision tree classifier

Abstract

This study introduces a weighted decision tree algorithm for prediction of graft survival in renal transplantation using preoperative patient's data. The objective was to identify the preoperative attributes that affect the graft survival. Between the years 2000-2009, renal allotransplantation was carried out for 889 patients at Urology and Nephrology Center which is the subject matter of this study. The ID3 algorithm was chosen to build up the decision tree using the weka machine learning software. A modification was made on ID3 to refine the results. A weighted vector was introduced. The element of such a vector represents the weight of each attribute which was obtained by trial and error. The results indicated that the weighted algorithm was successful in predicting the graft survival after one year and identifying the attributes affecting graft survival. Keywords: Decision Tree, Data Mining, ID3 Algorithm, Graft Survival, Kidney Transplantation.