Testing Goodness of Fit for Proportional Hazards Model with Censored Observations
针对随机删失数据，提出一种基于Cox偏似然的数值型拟合优度检验，无需虚拟时变协变量或时间轴划分，并证明了检验的一致性。
Abstract A numerical omnibus test of fit for the two-sample proportional hazards model is proposed for randomly censored observations. The test is derived from Cox's partial likelihood and does not need a dummy time-dependent covariate or any partition of the time-axis. Consistency of the test is established. Examples are provided for illustration.