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存活機率

Abstract

在許多臨床試驗與醫學研究中，每個研究對象的疾病演進過程會在追蹤過程中被監測。通常在研究進行中，除了個體是否發生死亡外，疾病惡化過程的訊息也會被收集，其中包括是否發生與演進過程有關的事件或生物標誌。在臨床上常將疾病惡化過程的訊息作為隨時間變動的解釋變數，利用對比風險模式描述疾病惡化過程對個體存活機率的影響。在本研究中將推廣Xu和O’Quigley的方法，在具時間變動解釋變數的對比風險模式下，利用疾病惡化過程過去的訊息預測後續的存活機率。最後，藉由模擬評估本研究提出之方法的表現。

In many clinical trials and medical studies, the course of disease for each individual is monitored during the follow-up period. Information of disease progression including the occurrence of events and biological markers associated with the development of disease as well as its death is often collected in the study. Proportional hazards models incorporating the information of disease progression as time-dependent covariates are frequently used to investigate the effect of disease progression on survival. Here we extend Xu and O’Quigley’s approach to predict the probability of subsequent survival given the past information of disease progression under such time-dependent covariate model. The performance of proposed approach is evaluated by a simulation study.