Sure Joint Screening for High Dimensional Cox's Proportional Hazards Model Under the Case-Cohort Design

Area of Science:

• Biostatistics
• Genomics
• Survival Analysis

Background:

• Case-cohort studies are efficient for rare outcomes but face challenges with ultrahigh-dimensional covariates.
• Identifying relevant genetic markers is crucial for understanding disease risk and developing targeted therapies.

Purpose of the Study:

• To develop a sure joint feature screening method for case-cohort designs with ultrahigh-dimensional covariates.
• To address the challenge of selecting truly important genetic predictors in large-scale genomic studies.

Main Methods:

• A sparsity-restricted Cox proportional hazards model is employed.
• An iterative reweighted hard thresholding algorithm approximates the joint screening estimator.
• The sure screening property is rigorously proven, ensuring high probability of retaining all relevant covariates.

Main Results:

• The proposed method demonstrates superior screening performance compared to existing methods for case-cohort designs.
• Effective identification of jointly correlated yet marginally uncorrelated covariates with event time.
• Successful application to high-dimensional genomic data from a breast cancer study.

Conclusions:

• The developed sure joint feature screening method is effective for ultrahigh-dimensional case-cohort studies.
• This approach enhances the ability to identify significant genetic covariates, particularly in complex correlation structures.
• The method and its MATLAB implementation are publicly available for broader research use.