Five Years Survival of Patients After Liver Transplantation and Its Effective Factors by Neural Network and Cox Poroportional Hazard Regression Models.

摘要: Background: Transplantation is the only treatment for patients with liver failure. Since therapy imposes high expenses to and community, identification of effective factors on survival such after transplantation valuable. Objectives: The current study attempted model (two years old above) using neural network Cox Proportional Hazards (Cox PH) regression models. event defined as death due complications transplantation. Patients Methods: In a historical cohort study, clinical findings 1168 who underwent transplant surgery (from March 2008 march 2013) at Shiraz Namazee Hospital Organ Center, Shiraz, Southern Iran, were used. To one five patients, PH accompanied by three layers feed forward artificial (ANN) method applied data separately their prediction accuracy was compared area under receiver operating characteristic curve (ROC). Furthermore, Kaplan-Meier used estimate probabilities in different years. Results: estimated probability 91%, 89%, 85%, 84%, 83%, respectively. areas ROC 86.4% 80.7% ANN models, addition, rate methods equally 92.73%. Conclusions: present detected more accurate results those analyze order patients’ clinically acceptable. large dataset few missing advantage this fact which makes reliable.