Variable selection for high-dimensional quadratic Cox model with application to Alzheimer's disease

Area of Science:

• Biostatistics
• Statistical modeling
• Genomics

Background:

• High-dimensional data presents challenges for traditional Cox models.
• Existing variable selection methods often fail with hierarchical structures.
• Quadratic Cox models are crucial for analyzing complex biological data.

Purpose of the Study:

• To develop a novel variable selection method for high-dimensional quadratic Cox models.
• To address the limitations of existing methods in handling hierarchical structures.
• To apply the method to a real-world Alzheimer's Disease study.

Main Methods:

• A penalized log partial likelihood approach is proposed.
• Generalization of the Regularization Algorithm under Marginality Principle (RAMP) for linear models.
• Extensive simulation studies to evaluate performance.

Main Results:

• The proposed method effectively performs variable selection in high-dimensional quadratic Cox models.
• The method successfully incorporates hierarchical model structures.
• Simulations demonstrate the method's practical utility.

Conclusions:

• The new penalized log partial likelihood approach offers a robust solution for variable selection in complex Cox models.
• This method is applicable to high-dimensional data with hierarchical structures.
• The approach shows potential for advancing research in areas like Alzheimer's Disease.