Post-selection inference for the Cox model with interval-censored data.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Inference

Background:

• Cox proportional hazards models are widely used for survival data analysis.
• Interval-censored data presents unique challenges in statistical modeling.
• Model selection methods like LASSO can introduce bias in subsequent inference.

Purpose of the Study:

• To develop a post-selection inference method for Cox models with interval-censored data.
• To provide asymptotically valid p-values and confidence intervals.
• To address challenges in statistical inference following LASSO model selection.

Main Methods:

• A novel post-selection inference method was developed for the Cox proportional hazards model.
• The method utilizes a pivotal quantity converging to a uniform distribution.
• Estimation of the efficient information matrix with consistent approaches was incorporated.

Main Results:

• The proposed method yields asymptotically valid p-values and confidence intervals.
• Simulation studies demonstrated satisfactory performance in modest sample sizes.
• The method's utility was confirmed in an Alzheimer's disease study.

Conclusions:

• The developed method offers reliable statistical inference for Cox models with interval-censored data after LASSO selection.
• This approach enhances the validity of results in complex survival data analyses.
• The method is applicable to real-world studies, including biomedical research like Alzheimer's disease.