Extended Cox model for breast cancer survival data using Bayesian approach: A case study

Abstract

Breast cancer (carsinoma mammae) is one type of cancer that occurs due to abnormal breast cell growth. Some of the factors that are thought to trigger breast cancer include the unhealthy lifestyles. The existence of these factors indicates that there is a correlation between breast cancer and patient survival. One of method for analyzing survival data is Cox proportional hazard. Cox proportional hazard model implies that each covariate is proportional. But in reality, there are often cases where there is a disproportionate covariate, in the sense that there is a relationship with the time, called time dependent covariate. In this case an extended of the Cox proportional hazard model needs to be done. Therefore, the aim of this paper to determine the relationship between the breast cancer patients' survival time and the factors that influence it using extended Cox model with Bayesian approach. This methodology is applied to breast cancer survival data from Hasanuddin University hospital in Makassar, Indonesia, for the period 2005-2018. The result shows the factors that substantially affect the breast cancer patients' survival time are marital status, histology, and leukocyte levels.