Oracle inequalities for the lasso in the Cox model

Abstract

We study the absolute penalized maximum partial likelihood estimator in sparse, high-dimensional Cox proportional hazards regression models where the number of time-dependent covariates can be larger than the sample size. We establish oracle inequalities based on natural extensions of the compatibility and cone invertibility factors of the Hessian matrix at the true regression coefficients. Similar results based on an extension of the restricted eigenvalue can be also proved by our method. However, the presented oracle inequalities are sharper since the compatibility and cone invertibility factors are always greater than the corresponding restricted eigenvalue. In the Cox regression model, the Hessian matrix is based on time-dependent covariates in censored risk sets, so that the compatibility and cone invertibility factors, and the restricted eigenvalue as well, are random variables even when they are evaluated for the Hessian at the true regression coefficients. Under mild conditions, we prove that these quantities are bounded from below by positive constants for time-dependent covariates, including cases where the number of covariates is of greater order than the sample size. Consequently, the compatibility and cone invertibility factors can be treated as positive constants in our oracle inequalities.

BFLCRM: A Bayesian functional linear Cox regression model for predicting time to conversion to Alzheimer’s disease

“…Third, BFLCRM can be regarded as an important extension of high-dimensional survival models. However, most high-dimensional survival models focus on the identification of a small set of covariates and their overall effect on time-to-event outcomes [5, 39, 43]. These approaches can be sub-optimal for high-dimensional imaging data, since the effect of imaging data on clinical data and other imaging data is often non-sparse , which makes it notoriously difficult for many existing regularization methods [20, 74].…”

Pseudo value method for ultra high-dimensional semiparametric models with life-time data

“…For Cox model, we can take a large constant with respect to λ to satisfy P ( β − β 1 > λ) = o(1) according to Huang et al (2013) for the LASSO penalty. For general trans-formation models, by Klaassen et al (2017), LASSO can also provide the desired result in (C2)…”

Adaptive kernel estimation of the baseline function in the Cox model with high-dimensional covariates

“…Assumption 3.2 is a classical assumption in the Cox model to obtain oracle inequalities (see Huang et al [17] and Bradic and Song [5]) and seems reasonable since in practice the covariates are bounded.…”

“…Let us introduce the centered process ν n,h (f ) = ⟨ᾱ h − α h , f ⟩ 2 , for any h ∈ H n and f ∈ L 2 ([0, τ ]) and B τ = {f ∈ L 2 ([0, τ ]), ∥f ∥ 2 = 1}. Under the assumptions of Theorem 4.1, with V (h ′ ) defined by (17) for any h ′ ∈ H n , there exist two constants c 6 and c 7 depending on the bound κ b of the Bürkholder Inequality, τ , ∥α 0 ∥ ∞,τ , the bound c S of…”