Bayesian semiparametric models for survival data with a cure fraction.

Area of Science:

• Biostatistics
• Survival Analysis
• Statistical Modeling

Background:

• Semiparametric survival models are essential for analyzing time-to-event data.
• Cure fraction models are necessary when a portion of the population is assumed to be immune to the event.
• Existing models may lack flexibility in capturing complex survival dynamics.

Purpose of the Study:

• To develop and validate a new class of semiparametric survival models incorporating a cure fraction.
• To introduce a smoothing parameter to control the degree of parametricity in the right tail of the survival distribution.
• To propose and investigate novel improper noninformative and informative prior classes for Bayesian inference.

Main Methods:

• Bayesian inference techniques applied to semiparametric cure rate models.
• Development of a smoothing parameter to enhance model flexibility.
• Derivation of novel theoretical properties for the proposed priors and posterior distributions.
• Application of the methodology to a melanoma clinical trial dataset.

Main Results:

• The proposed smoothing parameter is demonstrated to be crucial for accurate posterior estimates.
• Novel properties of the semiparametric cure rate model are derived.
• The behavior of posterior distributions under both noninformative and informative priors is examined.
• The methodology's practical utility is illustrated through a real-world clinical trial case study.

Conclusions:

• The developed Bayesian semiparametric survival models with a cure fraction offer a flexible and robust analytical framework.
• The smoothing parameter plays a critical role in refining survival distribution estimations.
• The proposed prior structures provide valuable tools for Bayesian analysis in survival data, especially with historical information.