• Online Journals
  • 21
specialty journal of medical research and health science
Volume 4, 2019, Issue 1
Analyzing the Survival of Patients with Liver Cirrhosis Using Ridge Semiparametric Models
Jamileh Abolghasemi, Shahnaz Rimaz, Mohsen Nasiri Toosi
Pages: 37-45


Introduction and Purpose of the study: Liver cirrhosis is one of the most common causes of death due to gastrointestinal diseases. It accounts for more than one million deaths annually in the world. Liver transplantation is the only way to cure this disease. Given the limited number of donated livers, the prioritization of the patients waiting for the transplantation queue is mandatory. The aim of this study was to analyze the survival of patients who were waiting for the liver transplantation using semi-parametric ridge regression models. Materials and Methodology: This study was a survival research. The data were collected from 305 patients waiting for the liver transplantation that were followed up at least for 7 years. Due to the correlation between the covariates, the ridge semiparametric models were used. Data analysis was performed using R (version 3.2.3) software. Results: In this study, out of 305 patients, 71 (23.3%) patients died because of liver cirrhosis and 51 patients (16.7%) had liver transplantation. The one-year, three-year, and five-year survival of the patients was 0.789, 0.556, and 0.478, respectively. In studying the factors affecting the survival of patients with liver cirrhosis, the variables of albumin logarithm, bilirubin log, age and encephalopathy were found to be statistically significant at the significance level of 0.01 in ridge regression models and Cox proportional hazards model. Using CVL indices, bias and total squared errors, the ridge regression model had better fit than Cox proportional hazards model. Conclusion: Due to the existence of a collinearity between the laboratory variables under study, using a ridge regression model for survival analysis of patients with liver cirrhosis reduced the estimation error and the bias of the fittings. Therefore, it is suggested that in the use of survival models, the collinearity between the covariates should be considered and, if any, some appropriate models should be used for the reduction of bias.

Call For Papers


[email protected]

specialty journal of medical research and health science
Issue 1, Volume 5, 2020