Deep survival forests with feature screening. (January 2023)

• Record Type: Journal Article
• Authors: Cheng, Xuewei; Wang, Sizheng; Zou, Yi; Abba, Badamasi; Wang, Hong
• Is Part Of: Biomedical signal processing and control. Volume 79(2023)Part 2
• Publication Date: 2023-01
• Subjects: Deep survival forests -- High-dimensional data -- Feature screening -- Error-screening -- Max-pooling; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2022.104195
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract: Deep forest has recently received much attention, mainly from the machine learning community, because it has fewer hyper-parameters, better performance, and is easier to construct than deep neural networks. This approach, as well as its variations, are commonly used for image classification, face recognition, music classification, and other related tasks. Its potential in survival analysis, on the other hand, has yet to be determined. In this paper, we propose a deep survival forests framework for modeling right-censored data with high and ultra-high dimensional covariates. We construct the proposed framework by merging the cascade survival forest structure with a feature screening mechanism. Experimental and statistical analysis results on well-known ultra-high and high-dimensional datasets show that the proposed methodology outperforms popular machine learning-based methods such as deep neural survival networks, random survival forests and their several variants in terms of prediction accuracy. Highlights: We try to explore non-NN (Neural Networks) style deep learning in the field of survival analysis. DSFfs is a deep model based on random survival forests, to model high-dimensional survival data. Deep survival forests provides an option when deep survival and other popular models are inferior.