Bayesian optimal response-adaptive design for binary responses using stopping rule.

Area of Science:

• Clinical Trials Methodology
• Bayesian Statistics
• Biomedical Research

Background:

• Response-adaptive designs enhance phase III clinical trials by assigning more patients to effective treatments.
• Current optimal designs primarily use a frequentist approach.
• There is a need for Bayesian methods in adaptive trial design.

Purpose of the Study:

• To propose novel response-adaptive designs for two-treatment phase III clinical trials using Bayesian prediction.
• To evaluate the properties and performance of these proposed designs.
• To demonstrate the practical application of the Bayesian methodology with real-world data.

Main Methods:

• Development of Bayesian prediction-based response-adaptive designs for two treatments.
• Theoretical analysis of the proposed designs' properties.
• Numerical comparisons with existing frequentist adaptive designs.
• Application and redesign of an experiment using a real clinical data set.

Main Results:

• The proposed Bayesian response-adaptive designs show promising properties for phase III trials.
• Numerical simulations indicate competitive or superior performance compared to existing methods.
• The methodology is applicable and effective when demonstrated on a real data set.

Conclusions:

• Bayesian prediction offers a viable and effective approach for developing response-adaptive designs in phase III clinical trials.
• The proposed methods provide a valuable alternative to frequentist designs, enhancing patient allocation to better treatments.
• The study validates the practical utility of Bayesian adaptive designs through real-world data application.