Comparison of survival models to study determinants liver cancer

Abstract

Cancer is one of the dangerous diseases that afflict a person through injury to cells and tissues in the body, where a person is vulnerable to infection in any age group, and it is not easy to control and multiply between cells and spread to the body. In spite of the great progress in medical studies interested in this aspect, the options for those with this disease are few and difficult, as they require significant financial costs for health services and for treatment that is difficult to provide.

This study dealt with the determinants of liver cancer by relying on the data of cancerous tumours taken from the Iraqi Center for Oncology in the Ministry of Health 2017. Survival analysis has been used as a method for analyzing the study data, because it applies to the phenomenon of interest, by comparing the parameter and semi-parametric models using a standardized (BIC, AIC) to determine the most appropriate model for data analysis, and the study concluded that the Cox parametric model with the lowest values for the evaluation criteria is the most appropriate for the analysis of the study data and from it, the most important variables with a significant effect on liver cancer were identified as Age and method of diagnosis and accurate diagnosis of the disease were significant variables affecting liver cancer. The study also found the significance of the relative risk for each of the variables (age, method of diagnosis, disease location, accurate diagnosis, the extent of prevalence) i.e. increased risk of liver cancer when these variables increase, each according to its specificity