以標誌時間為預測因子的條件存活機率估計

Estimation of Conditional Survival Probabilities Using Marker Time as Predictors

彭文君(Wen-Chun Peng)

指導教授 ： 張淑惠

國立臺灣大學/公共衛生學院/流行病學研究所/碩士(2010年)

https://doi.org/10.6342/NTU.2010.01345

摘要

在多階段疾病過程的發展中，在死亡之前和時間有關的標誌通常在預測基於疾病為有設限的三階段資料的存活上扮演了重要的角色。在這篇論文中，我們考慮將標誌時間當作預後因子來預測後續的存活。Cox模式是在存活資料中最普及的迴歸模式，為了研究預後因子及死亡時間的風險函數的關係。因此利用結合各種標誌時間的函數當作Cox模式的變數，在給定標誌時間的條件下的Cox模式及估計所對應的條件存活機率，並以一個實際例子來說明。

關鍵字

標誌狀態 ； 標誌時間 ； 存活機率 ； 預測

並列摘要

In the progression of a multistate disease process, times to a certain markers before death usually play important roles in predicting the survival of the disease based on the three-state data subject to censoring. In this paper, we consider marker times as prognostic factors to predict the subsequent survival. Cox models are the most popular regression models for survival data in order to investigate the relationship between prognostic faxtors and hazard function for death time. Therefore, Cox’s models are used by incorporating various functions of the marker times as covariates. The Cox’s model and the estimation of the corresponding conditional survival probabilities given marker time are provided. A real example is presented for illustration.

並列關鍵字

marker process ； marker time ； survival probability ； predict

參考文獻

Casella G, Berger R L., Statistical Inference, second edition. Duxbury 2002.

Dancourt V, Quantin C, Abrahamowicz M, Binquet C, Alioum A, Faivre J. Modeling recurrence in colorectal cancer. Journal of Clinical Epidemiology 2004; 57: 243-251.

Neri A, Marrelli D, Rossi S, De Stefano A, Mariani F, De Marco G, Caruso S, Corso G, Cioppa T, Pinto E, Roviello F. Breast cancer local recurrence: Risk factors and prognostic relevance of early time to recurrence. World Journal of Surgery 2007; 31: 36-45.

Xu RH, O'Quigley J. Proportional hazards estimate of the conditional survival function. Journal of the Royal Statistical Society Series B-Statistical Methodology 2000; 62: 667-680.

楊竺諺 (2009) 以疾病惡化過程為具時間變動解釋變數的對比風險模式預測後續