Statistical inference for response adaptive randomization procedures with adjusted optimal allocation proportions.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Statistical Inference

Background:

• Seamless phase II/III clinical trials are gaining traction, primarily utilizing Bayesian response adaptive randomization (RAR).
• Frequentist RAR designs for seamless trials are under-researched due to challenges in statistical inference.
• Existing frequentist RAR designs offer theoretical optimal allocation and asymptotic results.

Purpose of the Study:

• To investigate the asymptotic properties of frequentist RAR designs with adjusted target allocation proportions.
• To develop valid statistical inference methods for these frequentist seamless trial designs.
• To establish a theoretical foundation for advanced seamless clinical trials.

Main Methods:

• Studying asymptotic properties of frequentist RAR designs.
• Developing and investigating statistical inference procedures for these designs.
• Utilizing adjusted target allocation proportions.

Main Results:

• The proposed frequentist RAR design demonstrates desirable asymptotic properties.
• The statistical inference methods are valid for this procedure.
• Numerical studies confirm the design's ethical and efficient nature.

Conclusions:

• The developed frequentist RAR design offers a robust framework for seamless clinical trials.
• This research addresses a critical gap in frequentist methodology for adaptive trial designs.
• The findings support the implementation of ethical and efficient frequentist seamless trials.