Landmark Prediction of Survival Incorporating Short-Term Event Information

• Activity Number: 227
• Type: Topic Contributed
• Date/Time: Monday, August 1, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Layla Parast*+ and Su-Chun Cheng and Tianxi Cai
• Companies: Harvard University and Dana Farber Cancer Institute and Harvard University
• Keywords: survival analysis ; risk prediction ; time-varying coefficient model

Abstract

In recent years, genetic and biological markers have been examined extensively for their potential to signal progression or risk of disease. In addition to these markers, it has often been argued that short term outcomes may be helpful in making a better prediction of disease outcomes in clinical practice. Due to the potential difference in the underlying disease process, patients who have experienced a short term event of interest may have very different long term clinical outcomes from the general patient population. Most existing methods for incorporating censored short term event information in predicting long term survival focus on modeling the disease process and are derived under parametric models in a multi-state survival setting. In this paper we propose to incorporate short term event time information up to a landmark point along with baseline covariates via a flexible varying-coefficient model. The performance of the resulting landmark prediction rule is evaluated non-parametrically and compared to prediction rules constructed using the baseline covariates only. Simulation studies and an example suggest that the proposed procedures perform well in finite samples.