Date of Award


Degree Type


Degree Name

Master of Science




Dr. David Bellhouse


This study evaluated the performance of prognostic models for survival after liver transplant. Assessment of the adequacy of such models is difficult and quantitative benchmarks are needed for measuring performance. We examined the commonly used Cox proportional hazards (PH) model for survival analysis and compared to simpler models using survival trees. Models were evaluated using the integrated Brier score on an independent test set, allowing comparison of models based on prediction error. We also evaluated Harrell’s concordance statistic in the Cox PH model. We found that two important predictors of survival violated the PH assumption, suggesting that both the PH model and the concordance statistic are inappropriate for transplant data. We found that the scientific significance of the predictive accuracy gained through the use of the models tested here was limited. Benchmarks for performance evaluation are an important tool for accurate decision making in medicine.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.