Time-dependent ROC analysis for a three-class prognostic with application to kidney transplantation
Y Foucher1, M Giral, J P Soulillou
1Institute of Transplantation and Research in Transplantation, Nantes University, 30 bd J. Monnet, Nantes 44093, France. Yohann.Foucher@univ-nantes.fr
This study extends receiver operating characteristic (ROC) curve analysis for survival data to predict multiple competing risks. The new method enhances diagnostic accuracy assessment for markers predicting distinct, simultaneous events.
Area of Science:
- Biostatistics
- Medical Decision Making
- Survival Analysis
Background:
- Receiver operating characteristic (ROC) graphs are widely used in medical decision-making for diagnostic testing.
- Existing time-dependent ROC methods evaluate marker accuracy for predicting a single time-dependent failure.
- Analyzing multiple competing failure events, common in clinical practice, remains a challenge for current ROC methodologies.
Purpose of the Study:
- To extend the time-dependent ROC method to analyze the accuracy of a marker in predicting two competitive events.
- To provide a statistical framework for evaluating diagnostic marker utility when multiple outcomes are possible.
Main Methods:
- The study proposes an extension of the time-dependent ROC methodology.
- This extension is designed to handle scenarios with multiple, competing failure events.
- The approach is applied to evaluate marker accuracy for predicting distinct outcomes simultaneously.
Main Results:
- The proposed extension allows for the analysis of marker accuracy in predicting competitive events.
- This method addresses the limitation of previous ROC approaches that focused on single failure events.
- Demonstrates the utility of the extended ROC method in a medical context, such as kidney transplantation.
Conclusions:
- The extended time-dependent ROC method offers a valuable tool for assessing diagnostic marker accuracy in the presence of competing risks.
- This advancement improves the application of ROC analysis in complex medical situations with multiple potential outcomes.
- Facilitates more comprehensive evaluation of prognostic markers in survival data analysis.
Related Concept Videos
Receiver Operating Characteristic Plot
Comparing the Survival Analysis of Two or More Groups
Kidney Transplant I: Introduction
Acute Kidney Injury IV: Diagnostic Studies and Prevention
Acute Kidney Injury I: Introduction
Kidney Transplant II: Surgical Procedure
