A Comparative Study of Parametric and Non- Parametric Survival Analysis Models for Breast Cancer Patients

Abstract

The study dealt with the problem of determining an appropriate survival model for breast cancer patients data based on the distribution of survival time, where breast cancer is one of the most common causes of cancer death.This study aimed to find out the factors affecting on the survival analysis of breast cancer patients and also identify the effect of the distribution of survival time on the parametric survival models, also to determine the appropriate model among the parametric survival models and non-parametric (Cox model) by using (AIC( and )BIC) criteria. A sample of 246 patients was taken from Khartoum Oncology Hospital, 150 (61%) died, Survival times were simulated according to the distributions of the three studied models (Exponential-Weibull-Gompertz). The hypotheses of study is there an effect by the factors of the study on the survival time, and are there any differences between parametric models according to it is survival time distribution, and between them and Cox model. The results study found that the factors: education level, marital status, number of children, stage of disease and treatments (radiation-chemical-hormonalsurgical) have an effect on survival, While factors: age, function, tumor side are not an effect on survival time, and distribution of survival time effects on the appropriate parametric model, the Weibull model is the appropriate model based on the (AIC and BIC) criteria as it has the lowest value for both criteria. The study Recommended that to study the effect of the estimation methods of Cox model on the significance of the estimated parameters model, also to studying other parametric survival models and comparing them with each other and with the Cox model, concern to factors affecting survival and benefit from them in increasing the likelihood of survival as well as in prevention.