Predictive Assessment and Comparison of Bayesian Survival Models for Cancer Recurrence

2026-01-04

AI Summary

This study addresses the challenge of effectively evaluating and comparing Bayesian survival models in complex scenarios involving unmodeled censoring events and time-dependent covariates, which commonly arise in cancer recurrence data. The authors propose a systematic framework for predictive assessment and model comparison tailored to such settings, integrating Bayesian survival analysis, posterior predictive checks, and formal model selection criteria. Implemented through open-source software, this approach provides the first dedicated evaluation strategy for Bayesian survival models that account for both censoring and time-varying effects. By establishing a reproducible analytical pipeline, the framework significantly enhances the practical utility and reliability of these models in predicting cancer recurrence.

Technology Category

• Reasoning under Uncertainty: Relational Probabilistic Models
• Machine Learning: Bayesian Learning
• Cognitive Modeling & Cognitive Systems: Other Foundations of Cognitive Modeling & Systems

Application Category

• User Modeling, Personalization and Recommendation: User model development and evaluation
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Web Mining and Content Analysis: Models for Web evolution

Abstract

Complex data features, such as unmodelled censored event times and variables with time-dependent effects, are common in cancer recurrence studies and pose challenges for Bayesian survival modelling. Current methodologies for predictive model checking and comparison often fail to adequately address these features. This paper bridges that gap by introducing new, targeted recommendations for predictive assessment and comparison of Bayesian survival models. Our recommendations cover a variety of different scenarios and models. Accompanying code together with our implementations to open source software help in replicating the results and applying our recommendations in practice.

Problem

• Bayesian survival models
• cancer recurrence
• censored event times
• time-dependent effects
• predictive model comparison

Innovation

• Bayesian survival models
• predictive assessment
• cancer recurrence
• time-dependent effects
• censored data