useR! 2024

Innovative clinical trial designs typically involve advanced statistical methods and extensive simulations. Building these complex simulation pipelines introduces challenges in reproducibility and transparency not easily addressed by traditional development workflows. In this session we will present how to use the targets R package, a Make-like pipeline tool, to develop efficient and reproducible simulation pipelines for innovative clinical trial designs. We will then show how Git and GitHub Actions can be used to deploy these large-scale simulation pipelines to cloud computing instances/clusters. The combination of these tools results in a robust and efficient workflow, enhancing the reproducibility of complex simulation pipelines. We will provide a detailed walkthrough of our approach, complete with practical examples and best practices, making it a valuable resource for statisticians and research software engineers working on innovative clinical trial designs and beyond.