Abstract:
|
For some cancers especially in the early diagnosis, the chance of having disease progression or death could be very small after a certain time point and a portion of patients could be considered as "cured". This phenomenal could be illustrated by the flat tail in an estimated K-M survival curve. In this setting, traditional distributions such as the exponential or the weibull distribution to generate survival data will not be appropriate. This paper will demonstrate alternative parametric and non-parametric approaches to generate cure survival data and illustrate how this could be applied in oncology studies.
|