Variable Selection for Nonlinear Cox Regression Model via Deep Learning

Variable selection problem for the nonlinear Cox regression model is considered. In survival analysis, one main objective to identify covariates that are associated with risk of experiencing event interest. The proportional hazard being used extensively in analysis studying relationship between times and covariates, where assumes covariate has a log-linear effect on function. However, this linearity assumption may not be satisfied practice. order extract representative subset features, various variable approaches have been proposed data under linear model. there exists little literature To break gap, we extend recently developed deep learning-based LassoNet data. Simulations provided demonstrate validity effectiveness method. Finally, apply methodology analyze real set diffuse large B-cell lymphoma.