An information ratio-based goodness-of-fit test for copula models on censored data

Area of Science:

• Statistics
• Survival Analysis
• Biostatistics

Background:

• Copula models are vital for multivariate censored data analysis.
• Existing goodness-of-fit tests are limited to complete or right-censored data.
• Interval-censored and recurrent events data lack formal copula model assessment.

Purpose of the Study:

• To develop a general goodness-of-fit (GOF) test for copula-based survival models.
• To address the research gap for interval-censored and recurrent events data.
• To provide a versatile test applicable to various copula families.

Main Methods:

• Development of a general GOF test utilizing the information ratio (IR).
• The test is applicable to parametric copula families (Archimedean, Gaussian, D-vine).
• Establishment of asymptotic properties for the test statistic.

Main Results:

• The proposed test statistic is easy to calculate and implement.
• Simulations demonstrate good type-I error control and adequate power.
• The method effectively distinguishes between different copula models in real datasets.

Conclusions:

• The developed GOF test successfully addresses limitations in existing methods.
• It provides a reliable tool for evaluating copula models with complex censoring.
• The approach enhances the rigor of survival data analysis using copulas.