Piantadosi S, Crowley J
Johns Hopkins Oncology Center, Baltimore, Maryland 21205, USA.
Biometrics. 1995 Mar;51(1):249-58.
Parametric survival functions are usually defined as explicit functions of time and covariates. However, consideration of some simple differential equations describing certain survival curves leads to a descriptive equation which cannot be explicitly solved for the survival function. Nevertheless, the resulting survival function has useful properties: it is a hybrid between the exponential and uniform distributions, it can be used computationally in the same fashion as other survival models, and the effect of covariates on model parameters can be incorporated in a straightforward manner. The derivation and an example of the use of this model are provided.
参数生存函数通常被定义为时间和协变量的显式函数。然而，考虑一些描述特定生存曲线的简单微分方程会导致一个无法明确求解出生存函数的描述性方程。尽管如此，所得的生存函数具有有用的性质：它是指数分布和均匀分布之间的混合分布，可以以与其他生存模型相同的方式进行计算，并且协变量对模型参数的影响可以以直接的方式纳入。本文提供了该模型的推导过程及使用示例。
