The L1/2 regularization approach for survival analysis in the accelerated failure time model

• Published in: Computers in biology and medicine Vol. 64; pp. 283 - 290
• Main Authors: Chai, Hua, Liang, Yong, Liu, Xiao-Ying
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2015
• Subjects: Accelerated failure time model; Internal Medicine; L1/2 penalty; Other; Regularization; Survival analysis; Variable selection; L1/2 penalty; Survival analysis; Accelerated failure time model; Regularization; Variable selection; L 1/2 penalty
• ISSN: 0010-4825; 1879-0534
• DOI: 10.1016/j.compbiomed.2014.09.002

The analysis of high-dimensional and low-sample size microarray data for survival analysis of cancer patients is an important problem. It is a huge challenge to select the significantly relevant bio-marks from microarray gene expression datasets, in which the number of genes is far more than the size of samples. In this article, we develop a robust prediction approach for survival time of patient by a L1/2 regularization estimator with the accelerated failure time (AFT) model. The L1/2 regularization could be seen as a typical delegate of Lq(0<q<1) regularization methods and it has shown many attractive features. In order to optimize the problem of the relevant gene selection in high-dimensional biological data, we implemented the L1/2 regularized AFT model by the coordinate descent algorithm with a renewed half thresholding operator. The results of the simulation experiment showed that we could obtain more accurate and sparse predictor for survival analysis by the L1/2 regularized AFT model compared with other L1 type regularization methods. The proposed procedures are applied to five real DNA microarray datasets to efficiently predict the survival time of patient based on a set of clinical prognostic factors and gene signatures. •We propose a L1/2 penalized accelerated failure time (AFT) model.•A coordinate descent algorithm with renewed L1/2 threshold is developed.•The L1/2 penalized AFT model is able to reduce the size of the predictor in practice.•The classifier based on the model is suitable for the high dimension biological data.