Testing Equality of Survival Distributions when the Population Marks are Missing

Area of Science:

• Biostatistics
• Survival Analysis
• Nonparametric Statistics

Background:

• Comparing survival distributions is crucial in medical research.
• Standard log-rank tests fail with missing population data for censored observations.
• Existing methods may lack statistical power or be inapplicable.

Purpose of the Study:

• Introduce a nonparametric approach for testing survival distribution equality with censored data and missing population marks.
• Address limitations of standard log-rank tests in such scenarios.
• Enhance statistical power in survival data analysis.

Main Methods:

• Propose imputing population marks for censored observations.
• Develop fractional at-risk sets for a modified log-rank test.
• Utilize simulation studies to evaluate test performance.
• Derive an asymptotic linear representation for the test statistic.

Main Results:

• The proposed imputed log-rank test demonstrates increased power compared to methods that discard censored data.
• Simulation results validate the performance of the new testing methodology.
• An asymptotic linear representation of the test statistic was successfully obtained.

Conclusions:

• The imputed log-rank test provides a powerful and applicable nonparametric method for survival data with missing population marks.
• This approach enhances statistical power in survival analysis.
• The methodology is validated through simulations and real-world data application.