CDISC-SDMX

Automation of CDISC outputs using SDMX

Project goals

The goal of this project is a proof-of-concept to create a model of a clinical trial analysis using SDMX and to transform that model into CDISC (ADaM and ARM) compliant outputs.

Research questions

Key research questions for this project are:

• Demonstrate how SDMX can be used to describe typical clinical trial analyses based on the CDISC ADaM Examples in Commonly used Statistical Analysis Methods
• Can a domain specific language (DSL) be created that will allow analyses to be specified in a sufficiently high-level declarative language that supports easy human review and validation, which can be programmatically converted to SDMX
• Can SDMX annotations be used to automate a mapping to CDISC ADaM datasets. A stretch goal is to explore if this mechanism could be used for creation of CDISC ARM metadata.

How to run the project

The easiest way to run the project is to create and account with gitpod.io

Runtime environment

• The SDTM-to-CDISC programming is implemented in Python 3
• Powershell Core (i.e. Pwsh running on Ubuntu) is the build environment
• The operating system is Ubuntu docker image that is running in Gitpod

The environment is detailed in the .gitpod.Dockerfile and .gitpod.yml files in the root of the repo

Directory structure

This repo directory structure has been created to be compatable with the Phuse Script Metadata Recommendations for Sharing

There are some additional directories that are required for this project.

The Phuse standard directories are:

• ./conf – store any configuration files
• ./data – datasets used by scripts
• ./lib - script libraries, macros or utility programs
• ./log – store the log files; this starts empty.
• ./pkg – store R or other language packages
• ./script – store script files
• ./output – store output files such as dataset, reports, graphs, etc; this starts empty.

Additional directories:

• ./ref - reference material including standards, lookup tables, etc.
• ./docs - documentation