Uploaded July 2026 | Updated September 2026, 2 weeks ago
This demo shows how Mediforce uses AI agents, automated validation, and human review to build traceable tables, figures, and listings from study metadata and clinical datasets.
Created for CDISC AI Innovation 2026, the workflow begins with uploaded USDM metadata and SDTM datasets. An AI agent identifies the TFLs required for the study, audits the plan for completeness and traceability, and presents it for review before continuing.
Once the plan is approved, the workflow creates ARS-aligned analysis and ADaM specifications, derives the ADaM datasets, generates Define-XML 2.1, and runs CDISC CORE conformance checks. It then produces the Analysis Results Data and rendered TFLs for another round of human review.
The final output includes an interactive traceability explorer that connects study objectives and endpoints to each deliverable, ADaM dataset, and SDTM source. Review feedback can also be captured as reusable lessons and submitted to the project repository, helping the workflow improve over time.
The video covers:
• Planning and auditing required TFLs from USDM metadata
• Building ARS-aligned analysis and ADaM specifications
• Deriving ADaM datasets from SDTM
• Generating Define-XML 2.1 and running CDISC CORE checks
• Producing Analysis Results Data and rendered TFLs
• Reviewing and approving outputs at defined checkpoints
• Tracing objectives and endpoints through TFLs, ADaM, and SDTM
• Capturing reviewer feedback for future workflow improvements
Mediforce is an open-source platform for orchestrating transparent, human-supervised AI workflows in regulated environments.
Learn more about Mediforce: mediforce.ai
Explore the project on GitHub: github.com/Appsilon/mediforce
#CDISC #ClinicalData #TFL #ADaM #SDTM #PharmaAI #OpenSource
This demo shows how Mediforce uses AI agents, automated validation, and human review to build traceable tables, figures, and listings from study metadata and clinical datasets.
Created for CDISC AI Innovation 2026, the workflow begins with uploaded USDM metadata and SDTM datasets. An AI agent identifies the TFLs required for the study, audits the plan for completeness and traceability, and presents it for review before continuing.
Once the plan is approved, the workflow creates ARS-aligned analysis and ADaM specifications, derives the ADaM datasets, generates Define-XML 2.1, and runs CDISC CORE conformance checks. It then produces the Analysis Results Data and rendered TFLs for another round of human review.
The final output includes an interactive traceability explorer that connects study objectives and endpoints to each deliverable, ADaM dataset, and SDTM source. Review feedback can also be captured as reusable lessons and submitted to the project repository, helping the workflow improve over time.
The video covers:
• Planning and auditing required TFLs from USDM metadata
• Building ARS-aligned analysis and ADaM specifications
• Deriving ADaM datasets from SDTM
• Generating Define-XML 2.1 and running CDISC CORE checks
• Producing Analysis Results Data and rendered TFLs
• Reviewing and approving outputs at defined checkpoints
• Tracing objectives and endpoints through TFLs, ADaM, and SDTM
• Capturing reviewer feedback for future workflow improvements
Mediforce is an open-source platform for orchestrating transparent, human-supervised AI workflows in regulated environments.
Learn more about Mediforce: mediforce.ai
Explore the project on GitHub: github.com/Appsilon/mediforce
#CDISC #ClinicalData #TFL #ADaM #SDTM #PharmaAI #OpenSource
