Investigation of acute rejection status BY data mining methods in patients with kidney transplantation

Objective: Renal transplantation is a surgical implantation of a kidney from live donors or a cadaver for patients with end-stage renal disease. Kidney transplant failure is associated with adverse outcomes for patients, so it is important to investigate, identify and control risk factors. Data mining is a discipline that extract implicit, unknown and meaningful conclusions from databases. In this study, a descriptive statistical analysis of demographic, clinical and genetic factors of renal transplant patients is presented, and data mining results to understand acute rejection in renal transplant patients were aimed. Method: In this study, the data from 155 patients and donors who underwent renal transplantation between 01/06/2016 and 01/06/2017 in Akdeniz University Hospital were evaluated retrospectively. The data set contains clinical, laboratory, genetic and demographic data for each patient and donor. SPSS, SimMine and Weka software packages were used to examine the relationship between features. Descriptive statistical methods, logistic regression and genetic algorithms from data mining methods were applied on the data. Results: Demographic data of renal transplant patients and donors were analyzed by descriptive statistical methods. Significant relationships were found in terms of clinical features. Genetic Algorithms method has been used to predict acute rejection. In addition, statistically significant relationships were determined in the Chi-Square analyzes performed on demographic, clinical and genetic factors. Conclusion: Demographic factors of renal transplant patients were presented in Akdeniz University Hospital. In addition, a rule was established for the prediction of acute rejection from serious complications that may develop in patients undergoing renal transplantation. Although the detection of serious complications in renal transplants is currently done with clinical evaluations, it is inevitable that existing or new data mining methods will become more important in this process.