Agile by adaptive design: An algorithm for decentralized trials

Decentralized Clinical Trials (DCTs) represent a significant advancement in clinical research, offering greater accessibility, flexibility, and participant engagement through the use of telemedicine, mobile health technologies, and remote data capture. However, the decentralized nature of data collection introduces challenges related to data reliability and variability, which are often inadequately addressed by conventional statistical methods at the design stage of the trial. This study presents an agile Bayesian design framework tailored to the specific needs of DCTs, integrating adaptive data reliability directly into trial design and analysis. Our approach is based on Bayesian decision rules to guide interim sample size adjustments. By treating data reliability as a model parameter rather than an external factor, our method accounts for uncertainty and improves the robustness of power calculations. Simulation studies demonstrate the effectiveness of this strategy. The proposed framework enables a flexible and agile approach to DCT design that can adapt to varying data quality conditions. This work offers a foundation for extending the proposed adaptive method to other trial contexts, including time-to-event endpoints, and supports the broader adoption of DCTs in real-world clinical research.