Challenge
The development of real-world evidence was constrained by fragmentation and lack of standardisation.
Key issues included:
- Incompatible data structures across systems
- Absence of unified data models
- Inefficient cohort identification workflows
- Lack of controlled analytical environments
Approach
A structured data environment was implemented to support consistent dataset construction and analysis.
This included:
- Alignment of datasets to common structural models
- Controlled environments for sensitive data processing
- Cohort discovery capabilities across datasets
- Reproducible analytical workflows
Impact
- Faster cohort development
- Improved reproducibility of analysis
- Increased consistency across studies
- Reduced reliance on manual data preparation
Perspective
Reproducibility is constrained more by data inconsistency than analytical capability.
Differences in data structure and transformation introduce variation that cannot be resolved at the analysis stage.
The focus must shift from analysis to data construction. Without consistent, traceable datasets, results are not comparable across studies.
Standards & Frameworks
Standards and governance frameworks were embedded into data models and analytical workflows to ensure reproducibility, consistency, and traceability of observational data analysis.
This included:
- OMOP Common Data Model consistent structure for observational datasets
- CDISC where applicable alignment with structured research standards
- FAIR Principles reproducibility and data reuse
- Analytical reproducibility frameworks versioned and traceable workflows
- Data governance frameworks controlled analytical environments
Interested in a similar initiative?
Open to discussions with institutions exploring governance-aligned collaboration, secure environments, or regulated innovation partnerships.