Prediction of 1-year Graft Survival Rates in Kidney Transplantation: A Bayesian Network Model

Abstract

A Bayesian network model was developed to predict 1-year graft survival rates after kidney transplantation. The data from a random sample of 90% of 35,366 kidney transplants performed in the United States between 1987 and 1991 were used to build a Bayesian network model. The discriminating power and the accuracy of the model predicting 1-year graft survival rates was demonstrated using the remaining 10% test sample. By more accurately predicting graft survival, such models may be used to refine existing rule-based transplant-allocation systems and, thus, improve patient and transplant outcomes.