Variable selection in semiparametric nonmixture cure model with interval-censored failure time data: An application...

Area of Science:

• Biostatistics
• Survival Analysis
• Medical Informatics

Background:

• Censored failure time data with cured subgroups are common in cancer screening and tumorigenicity studies.
• Analyzing such data is complicated by a large number of potential risk factors, necessitating effective variable selection.

Purpose of the Study:

• To develop a variable selection method for semiparametric nonmixture cure models with interval-censored data and a cured subgroup.
• To address the challenge of high-dimensional risk factors in survival data analysis.

Main Methods:

• Proposed a penalized likelihood approach using LASSO, ALASSO, or SCAD penalties.
• Developed a novel penalized expectation-maximization algorithm for efficient computation.
• Validated the methodology through extensive simulations and a real-world prostate cancer screening dataset.

Main Results:

• The penalized likelihood approach effectively performs variable selection in the presence of cured subgroups and interval-censored data.
• The proposed penalized expectation-maximization algorithm provides a computationally feasible solution.
• Demonstrated the practical utility of the method in analyzing prostate cancer screening data.

Conclusions:

• The developed variable selection method is effective for analyzing complex survival data with cured subgroups.
• This approach enhances the understanding of risk factors in diseases like prostate cancer.
• Offers a robust statistical tool for biostatistical and epidemiological research.