Analyzing Receiver Operating Characteristic Curves with SAS

ROC Curves with Censored Data

7.1 Introduction

7.2 Lung Cancer Example

7.3 ROC Curves with Censored Data

7.4 Concordance Probability with Censored Data

7.5 Concordance Probability and the Cox Model

7.1 Introduction

In many applications, the binary outcome event is not immediately observable. For example, most credit scoring algorithms try to predict the probability of default by a certain time. If every subject in the data set is under observation at least until that time, then the outcome is truly binary and the methods we have seen so far are applicable. But it may not be desirable to wait until the outcome for all subjects is observed. ...