The Elastic Commensurate Prior Model for Historical Information Borrowing
10.11783/j.issn.1002-3674.2025.05.002
- VernacularTitle:弹性相称先验模型及其在历史信息借用中的应用
- Author:
Jike HUANG
1
;
Zhiwei RONG
;
Jiali SONG
Author Information
1. 北京大学公共卫生学院生物统计学系(100191)
- Publication Type:Journal Article
- Keywords:
Elastic commensurate prior;
Commensurate prior;
Elastic prior;
Clinical trial;
Historical data;
Information borrowing
- From:
Chinese Journal of Health Statistics
2025;42(5):649-654
- CountryChina
- Language:Chinese
-
Abstract:
Objective The objective of this study was to construct a new Bayesian information borrowing model based on the concept of elastic prior,evaluate the statistical properties of the method through simulation studies,and provide a new method for historical information borrowing.Methods The concept of elastic prior was introduced into the commensurate prior method to establish the elastic commensurate prior model.Simulations were conducted for clinical study outcomes as normal variables under situations where historical data was either consistent or inconsistent with the current study data.Type Ⅰ error,statistical power,and the 95％posterior highest density credible interval were used as evaluation criteria to compare the performance of the elastic commensurate prior method with non-informative prior,full-information prior,power prior,and commensurate prior methods.Results Through simulation studies in various scenarios,it was shown that the proposed elastic commensurate prior method,compared to other prior methods under similar conditions,provides better control of Type Ⅰ error,higher statistical power,and relatively narrower posterior highest density credible intervals.Conclusion The proposed elastic commensurate prior method not only better controls Type Ⅰ error but also ensures higher statistical power,improving the accuracy of estimating treatment effects.This method introduces a new approach for borrowing information from historical data in clinical trials.
