透過您的圖書館登入
IP:3.144.36.141
  • 學位論文

貝氏對右設限資料勝算比模型之研究

Bayesian Survival Analysis for Proportional Odds Model with Right Censored Data

指導教授 : 吳裕振

摘要


存活分析的資料蒐集上, 經常會發生存活時間(survival time)或稱事件發生時間無法被完整記錄下來的情況。 舉個例子, 今天有一個研究想要知道某種藥物對於死亡率的影響, 假設這是一個為期十年的研究, 在這類型的研究中可能出現以下情況, (1)到觀察時間截止觀察個體依然活著 (2)觀察個體失去追蹤 (3)觀察個體主動退出研究, 在這種情況下,我們無法得知目標是否死亡, 換句話說因為研究停止(end of the study)或是因為失去追蹤(loss of following up)等等情形,使得事件發生時間無法被觀察到,我們稱之為右設限(right censored), 在此我們定義追蹤時間(following time) X=min{T,C},其中 $T$ 為存活時間;C 為設限時間(censoring time), 則右設限資料(right censored data)其觀察值僅包括追蹤時間 X,以及事件發生時間 $T$ 是否在設限時間 C 之前發生。 對於右設限資料,本篇論文選擇比例勝算比模型(Proportional Odds Model)進行討論(詳見 Pettiet(1982) 以及 Bennett(1983a,b)),在存活資料的分析上,討論共變量Z以及事件發生時間T(failure time)之間關係的半母迴歸方法(semiparametric regression)已被廣泛的被研究,例如比例風險模型(proportional hazards model),比例勝算比模型(proportional odds model),可加性風險模型(additive hazards model), 線性轉換模型(linear transformation model),加速失敗時間模型(accelerated failure time model)和治癒模型(cure model)。 本篇論文的內容:第二節介紹伯氏多項式之幾何性質, 第三節介紹資料型態以及模型, 第四節導出其概似函數並進行貝氏推論, 第五節為M-H Green演算法以及RJ-MH演算法的流程, 第六節為模擬計算, 第七節為討論以及未來研究方向。

並列摘要


The purpose of this paper is the proportional odds model in the right censored data to go on estimated parameters , that using Bayesian methodsand observe the its conformance. in survival analysis of the data collection, often survival time (survival time) or the situation of the incident could not be fully recorded. Because end of the study (end of the study) or lost of following up (loss of following up) case, the incident could not be observed, which we call right the censored (right the censored), then right the censored data (right the censored data) observations only include tracking ,but also time X and time T incident within a set limit before the time C. We believe that the proportional odds function as a kind of continuous function, the KM method estimates the status quo data survival function that only get a step function, and in the number of samples is great in order to estimated smooth curve, we proposed that the Bernstein polynomials to estimate, because Bernstein polynomial easy to take into account the geometric information, and a smaller number of samples to estimate a smooth curve. Of this paper: Section II describes the Bernstein polynomial coefficients and graphics. Section III presents the status quo data types. Section IV to derive the likelihood function and Bayesian inference. Section V for the algorithm. Section VI for the simulation.

參考文獻


model for the analysis of current status data" , Journal of the American Statistical Association 1996.
[1] A. J. Rossini , A. A. Tsiatis "A semiparametric proportional odds regression
[2] S. Bennett , "Analysis of Survival Data by Proportional Odds Model" Statistics Medicine 1983a "Log-Logistic Regression Models of Survival Data" Applied Statistics 1983b
[3] W. Q. Fang " Bayesian Survival Analysis for Current Status Data ". Department
[4] I.S. Chang , C.A. Hsiung , Y.J. Wu , C.C. Yang " Bayesian Survival Analysis

被引用紀錄


劉懋婷(2013)。使用伯氏多項式對右設限資料勝算比模型之最大概似估計〔碩士論文,中原大學〕。華藝線上圖書館。https://doi.org/10.6840/CYCU.2013.00431

延伸閱讀