Extended model-based designs for more complex dose-finding studies.

Area of Science:

• Biostatistics
• Clinical Trial Design
• Pharmacometrics

Background:

• Model-based designs are crucial for efficient dose-finding studies.
• The continual reassessment method (CRM) is a common model-based approach.
• Single-model approaches can limit flexibility in complex trial scenarios.

Purpose of the Study:

• To explore extensions of model-based designs for enhanced flexibility in dose-finding studies.
• To demonstrate how using multiple models improves adaptability over single-model designs.
• To address complexities such as subject heterogeneity, varied treatment schedules, and partial ordering.

Main Methods:

• Utilizing established results from Bayesian model choice for inferential simplicity.
• Implementing a multi-model framework instead of relying on a single model.
• Developing extended models to accommodate specific complexities in dose-finding.

Main Results:

• The multi-model approach significantly increases design and analysis flexibility.
• Bayesian model choice simplifies the inferential process.
• The framework effectively handles added complexities like subject heterogeneity, different treatment schedules, and partial ordering.

Conclusions:

• Extending model-based designs with multiple models offers greater flexibility for dose-finding studies.
• This approach simplifies inference through Bayesian model choice.
• The enhanced flexibility allows for the incorporation of diverse complexities in clinical trial design.