標題: 半競爭風險資料受限於左截切的半母數推論
Semi-parametric Inference for Semi-competing Risks Data subject to Left Truncation
作者: 林怡君
Yichun Lin
王維菁
Weijing Wang
統計學研究所
關鍵字: 半競爭風險資料;左截切;2×2 列聯表;設限;Archimedean copulas model;Left truncation;Semi-competing risks data;Two-by-two table
公開日期: 2006
摘要: 本論文考慮以半母數推論方法估計在半競爭風險資料受限於左截切下的關連性。半競爭風險是多重事件發生的過程。以糖尿病的例子做說明,觀察值在罹患糖尿病之後往往會伴隨著一些併發症,如腎臟病或眼睛的病變而導致死亡,併發症與死亡之間的關係通常為醫學研究者所感興趣的。然而一些研究時間的限制,使得一些觀察值無法進入研究而被觀察到,我們稱這樣的觀察值受到截切。本論文的推論方法就是合併這兩個資料結構而進行,我們回顧了一些相關文獻,其中Jiang等人在2005年以concordance方法對同樣的資料結構做推論。而我們所提出的方法是以觀察值建立一系列2×2列聯表,此法可以視為Clayton在1978年對二維設限資料所提出的條件概似估計法的延伸。concordance方法與2×2列聯表皆利用了log-rank的概念建立估計函數,我們將以模擬的結果比較concordance方法與我們所提出的方法。
The thesis considers semi-parametric inference for estimating the association parameter for a copula model based on semi-competing risks data which are further subject to left truncation. We review related literature including the paper by Jiang et al. (2005) who suggest solving the same problem by using concordant indicators. Alternatively we propose to construct an estimating function based on a series of 2×2 tables. Our method can be viewed as an extension of the conditional likelihood approach proposed by Clayton (1978) who originally considered bivariate censored data. Simulations are performed to assess the finite-sample performance.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT009426523
http://hdl.handle.net/11536/81463
Appears in Collections:Thesis


Files in This Item:

  1. 652301.pdf