Estimation of Prediction Error for Survival Time with Application in Colorectal Cancer

Colorectal cancer is one of the most widespread and killer among canc-ers and it is highly important to predict the survival time. The purpose of this paper is fitting the Cox model and Kaplan-Meier cu-rve for the colorectal cancers data in Tale-ghani Hospital and comparing both with apparent loss. This paper is related to colorectal cancer patients in Tehran Tal-eghani Hospital who were followed up for at least five years. These patients referred to Taleghani Hospital from 2001 to 2006 and their information was recorded in the cancer section of Digestion Research Center and was followed up for at least five years. The Cox Proportional Hazards model and Ka-plan- Meier curve were used for prediction of survival times and these were compared with estimation of prediction error by apparent loss method. Cox model showed a significant relation between size of tumor and sex with survival time. The prediction error with ap-parent loss method for Cox model was at times less than Kaplan-Meier curve. Discussion & In this study، we concluded that estimation of prediction error could be used for comparison of mo-dels