A Bayesian decision-theoretic sequential response-adaptive randomization design.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Medical Research Methodology

Background:

• Phase II clinical trials are crucial for evaluating treatment efficacy but often face challenges with efficiency and ethical patient allocation.
• Current designs may not optimally balance early stopping for futility/efficacy with adaptive patient assignment to better treatments.

Purpose of the Study:

• To develop and evaluate a new class of phase II clinical trial designs.
• To enhance trial efficiency and ethical considerations through sequential stopping and adaptive treatment allocation.
• To reduce average sample sizes and increase patient allocation to more effective treatments.

Main Methods:

• Bayesian decision-theoretic sequential approach combined with adaptive randomization procedures.
• Incorporation of decision costs into design parameters for practical application.
• Development of a constrained backward induction and forward simulation algorithm for implementation.

Main Results:

• Proposed designs allow for early stopping due to efficacy or futility.
• Adaptive randomization successfully allocates more patients to superior treatment arms.
• Simulated trials demonstrate desirable operating characteristics and robustness to control group response rates.

Conclusions:

• The novel designs offer a more efficient and ethical framework for phase II clinical trials.
• The developed algorithm makes these advanced designs computationally practicable.
• These adaptive, sequential designs represent a significant advancement in clinical trial methodology.