The comparisons of random survival forests and Cox regression analysis with simulation and an application related to breast cancer

Authors:
Imran Kurt Omurlu;Mevlut Ture;Füsun Tokatli
Affiliations:
Trakya University, Medical Faculty, Department of Biostatistics, 22030 Edirne, Turkey;Trakya University, Medical Faculty, Department of Biostatistics, 22030 Edirne, Turkey;Trakya University, Medical Faculty, Department of Radiation Oncology, Edirne, Turkey
Venue:
Expert Systems with Applications: An International Journal
Year:
2009

Citing 4
Cited 1

Applied Survival Analysis: Regression Modeling of Time to Event Data

Applied Survival Analysis: Regression Modeling of Time to Event Data
Random Forests

Machine Learning
On the exact distribution of maximally selected rank statistics

Computational Statistics & Data Analysis
Using Kaplan-Meier analysis together with decision tree methods (C&RT, CHAID, QUEST, C4.5 and ID3) in determining recurrence-free survival of breast cancer patients

Expert Systems with Applications: An International Journal

Estimation and optimisation of right-censored data in survival analysis by neural network

International Journal of Business Information Systems

Quantified Score

Hi-index	12.06

Visualization

Abstract

The objective of this study was to compare the performances of Cox regression analysis (CRA) and random survival forests (RSF) methods with simulation and a real data set related to breast cancer. In the simulations, we compared across the methods under varying sample sizes by using Monte Carlo simulation method. The results showed that the performance of the CRA was a slightly better for analysis based on Harrell's concordance index than RSF approaches based on log-rank, conservation of events, log-rank score and approximate log-rank splitting rules. In the real data application, a retrospective analysis was performed in 279 breast cancer patients diagnosed. According to Harrell's concordance index, RSF based on approximate log-rank splitting rule to determined major risk factors for disease-free survival (DFS) showed a slightly better performance than other approaches. In general, performances of all the methods were almost similar. The predictive capability of CRA can be used for different sample sizes and potential future suitable survival data problems, whereas RSF provide interpretive results.