擴充風險模型之效率估計本研究將嘗試找出擴充風險模型之效率估計。此一估計由概似函數的核心平滑化，再藉著數值方法例如一般的梯度法找到最佳估計。擴充風險模型包含了最常用的兩大存活模型如Cox 模型及加速失敗時間模型。在巢狀結構之下，擴充風險模型提供了一個可能的工具能夠檢查Cox 模型及加速失敗時間模型的適當性。我們提出的選模程序可由概似比檢定來達成。 Efficient estimator of semi-parametric extended hazard model This study attempt to propose an efficient estimation for an extended hazard (EH) model, which includes the Cox proportional hazard (PH) and the accelerated failure time (AFT) models as special cases. The efficient estimation is derived from a kernel-smoothed profile likelihood function, which is a continuous and monotone function such that optimization can be achieved by gradient method. Due to the nested structure of EH model, model selection between PH and AFT models can possibly be done through likelihood ratio test. 研究期間：9908 ~ 10007
