ABSTRACT
A cancer clinical trial is usually a planned experiment to study the survival of a group of cancer patients and make inferences regarding the general patient population. Each patient has a failure time after the patient is enrolled in the study. In general, the failure time of an individual is a nonnegative random variable, which follows the survival distribution of the population. This chapter introduces several widely used parametric survival distributions and summarizes their characteristics. It discusses non-parametric methods of estimating the survival distribution for censored survival data. The parametric survival distributions can be estimated using maximum likelihood methods. The survival distribution can be estimated non-parametrically from observed survival data by using the method of Kaplan and Meier. Since the distribution of survival times tends to be positively skewed, the median survival time is the preferred summary measure of the central location of the survival distribution.
