Translating Predictive Distributions into Informative Priors: Method and Examples

Andrew A. Manderson's 2026 paper presents a method for translating elicited prior information into informative joint priors for Bayesian models. The 11.1 MB PDF includes supplementary materials for reproducing the work and is licensed under CC-BY-4.0. The paper illustrates the approach with three examples: a cure-fraction survival model, a setting with prior information on R², and a nonlinear regression model.

Use Cases

• Implementing the described method to construct informative priors for cure-fraction survival models based on the described mixed discrete/continuous observable quantity.
• Applying the translation approach to incorporate prior beliefs about model-derived quantities like R² into a joint prior distribution.
• Using the global, multi-stage Bayesian optimization procedure to locate optimal hyperparameters for a parametric class of prior distributions.
• Reproducing the work's examples to validate the method for nonlinear regression models as described in the paper.

Strengths

• Includes supplementary materials with a standardized description for reproducibility.
• Published under a permissive CC-BY-4.0 license allowing for reuse and adaptation.
• Provides concrete examples across different modeling domains (survival analysis, regression).

Limitations

• Dataset row count is unknown, which may limit suitability assessment for direct data analysis.
• Column-level documentation is absent; field semantics must be inferred from the paper after download.
• The primary file format is PDF, which may require extraction or manual transcription of any underlying data.

Provenance

Source

figshare

Collection Method

Methodological paper with accompanying supplementary materials.

Time Range

2026

Freshness

Last updated 2026-04-22 15:09:17; freshness should be verified.

Geography

null