463 – SPEED: Statistics in Epidemiology and Genomics and Genetics
Subgroup Analysis of Censored Survival Data on Cancer Treatment
Qing Zhang
Stony Brook University
Hongshik Ahn
Stony Brook University
In this study, we develop a statistical method based on tree-structured classification to assign patients from different treatment groups into subgroups with different medical recommendations. Since it is difficult to discover treatments that benefit all patients, we want to identify subgroups of patients for whom the treatment has an enhanced effect. Our model is applied to survival data. We classify terminal nodes into subgroups by comparing the relative event rates of the terminal nodes and the relative event rates of their corresponding immediate predecessors. Given the suggested subgroup of each terminal node, we give a suggested method on how to identify which of the splitting variables are keen to which of the subgroups by tracing back along the tree. Performance of our proposed method is evaluated from multiple simulation runs. The result shows that our method is more likely to give an appropriate recommendation when the treatment effect is more heterogeneous.