Bayesian analysis of generalized odds-rate hazards models for survival data

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• The Cox proportional hazards model is widely used for censored survival data.
• The assumption of proportional hazards is often violated in real-world scenarios.
• This limitation necessitates alternative modeling approaches.

Purpose of the Study:

• To introduce and re-examine the theoretical and computational properties of generalized odds-rate regression models.
• To provide a flexible class of models accommodating nonproportional hazards.
• To demonstrate the utility of these models in survival data analysis.

Main Methods:

• Consideration of a generalized odds-rate class of regression models.
• Re-examination of theoretical and computational properties.
• Establishment of posterior propriety under mild conditions.
• Simulation studies and real-data analysis (prostate cancer).

Main Results:

• The generalized odds-rate models encompass common survival models like proportional hazards, proportional odds, and accelerated lifetime models.
• The proposed methodology is validated through simulation and a prostate cancer dataset analysis.
• The models offer a robust framework for handling nonproportional hazards in survival data.

Conclusions:

• Generalized odds-rate regression models provide a valuable extension for analyzing censored survival data, particularly when the proportional hazards assumption is violated.
• The re-examined properties and demonstrated applications confirm their utility and flexibility.
• This approach enhances the analysis of complex survival data in various scientific fields.