The generalized odd log-logistic-G regression with interval-censored survival data

Area of Science:

• Statistics
• Survival Analysis
• Biostatistics

Background:

• Interval-censored data presents unique challenges in survival analysis as exact event times are not observed.
• Existing lifetime distributions may not fully capture the complexities of data where events occur within intervals.
• The generalized odd log-logistic family offers a flexible framework for modeling various risk function shapes.

Purpose of the Study:

• To propose a new regression model based on the generalized odd log-logistic family for analyzing interval-censored survival data.
• To extend existing interval modeling capabilities by leveraging the properties of this generalized family.
• To provide robust methods for parameter estimation and model assessment.

Main Methods:

• Development of a regression framework utilizing the generalized odd log-logistic distribution.
• Application of both classical and Bayesian methodologies for parameter estimation.
• Evaluation of model performance through simulation studies varying sample sizes and censoring percentages.
• Assessment of goodness-of-fit using likelihood ratio tests, residual analysis, and graphical techniques.

Main Results:

• The proposed generalized odd log-logistic regression model demonstrates effectiveness in handling interval-censored data.
• Parameter estimates show stable behavior across different sample sizes and censoring levels.
• Goodness-of-fit diagnostics confirm the suitability of the proposed models.
• The model's utility is validated through application to two real-world datasets.

Conclusions:

• The generalized odd log-logistic regression model provides a valuable and flexible tool for survival analysis with interval-censored data.
• The proposed estimation and validation methods are robust and applicable to practical scenarios.
• This approach enhances the analysis of data where precise event times are unavailable, offering insights into survival patterns.