Agile by adaptive design: An algorithm for decentralized trials

Area of Science:

• Clinical Research Methodology
• Biostatistics
• Digital Health

Background:

• Decentralized Clinical Trials (DCTs) enhance accessibility and engagement using digital health technologies.
• Data reliability and variability in DCTs pose challenges for traditional statistical methods.
• Current trial designs often inadequately address data quality issues inherent in remote data collection.

Purpose of the Study:

• To present an agile Bayesian design framework specifically for DCTs.
• To integrate adaptive data reliability directly into trial design and statistical analysis.
• To enhance the robustness of power calculations and adapt to varying data quality.

Main Methods:

• Developed an agile Bayesian design framework incorporating adaptive data reliability.
• Utilized Bayesian decision rules for interim sample size adjustments.
• Treated data reliability as a model parameter to account for uncertainty.

Main Results:

• Simulation studies confirmed the effectiveness of the proposed adaptive strategy.
• The framework demonstrated flexibility in adapting to different data quality conditions.
• Improved robustness in power calculations by integrating data reliability into the model.

Conclusions:

• The proposed Bayesian framework offers an agile and flexible approach for DCT design.
• This method enhances the reliability and robustness of clinical trials conducted remotely.
• The framework provides a foundation for adapting adaptive methods to other trial types and endpoints.