Application of Special Parametric Methods to Model Survival Data

Abstract

This paper deals with some new forms of probability distributions that can be used for parametric survival analysis. A special transformation is introduced to transform standard distributions to more flexible ones that allow us to model bathtub hazard functions and prevent inaccuracies in estimates of survival and hazard functions in longer survival times. Some commonly used distributions are transformed to the new forms. Basic characteristics of these transformed distributions are described in details. The standard distributions and the new ones are used to model the survival function of the patients suffering from breast cancer. The results are compared according to Akaike’s information criterion and examples of the use of the new derived distributions are presented, as well.