Neutrosophic Cox Proportional Hazards Model for Robust Variable Selection in Survival Analysis

Abstract

This paper introduces a novel approach for variable selection in survival analysis by integrating neutrosophic logic into the Cox Proportional Hazards (Cox PH) model to address the limitations of recent studies related to high dimensionality. Neutrosophic logic, is a mathematical framework that allows for uncertainty, indeterminacy, and inconsistency, and particularly well suited for handling the complexity and often-ambiguous nature of biological data. By incorporating neutrosophic sets into the Cox PH model, we aim to enhance model robustness, improve variable selection, and address the curse of dimensionality. We compare the performance of the neutrosophic-enhanced Cox PH model with traditional variable selection methods using real-world gene expression data, focusing on breast cancer survival prediction. The analysis results of this type of data concluded that using neutrosophic logic handled the issue of high-dimensional and improved the model performance.