RESEARCH & SECONDARY USE CONNECTORS

Data Transformation, Pseudonymization & Secure Research Environments

5 of 6

Enable governed secondary use of health data for research, real-world evidence and analytics with privacy, provenance and controlled outputs.

TransformationPseudonymizationLineageData QualitySecure EnvironmentOMOPCDISC

Privacy by Design

Protect identifiers and minimize re-identification risk.

Standards-Based Transformation

Transform FHIR into approved research models with MSIS harmonization.

Secure by Default

Bring computation to controlled data environments.

Traceable & Reproducible

Record lineage, versioning and audit throughout.

End-to-End Transformation Pipeline

Source Data

Authorized clinical/public sources

MSIS Normalization

Terminology, deduplication and quality context

Transformation

OMOP/CDISC/study-specific model and variables

Pseudonymization

Tokenize/remove/generalize identifiers

Quality Validation

Completeness, conformance, plausibility and lineage

Secure Environment

Controlled analysis; restricted raw export

Results & Output

Aggregated/approved governed exports

Pseudonymization

Technique	Use	Control
Remove direct identifiers	Names/contact/direct IDs not needed	Data minimization
Tokenize identifiers	Stable linkage without direct identifier	Protected key/token service
Generalize dates/geography	Reduce uniqueness	Research-purpose risk balance
Suppress rare categories	Reduce disclosure risk	Output/privacy thresholds

Data Quality & Lineage

Source extraction/version provenance

MSIS mapping/terminology versions

Transformation rule/code version

Quality checks and exceptions

Pseudonymization policy/version

Dataset checksum/version and approved consumer

Secure Environment

Controlled Access

Approved researcher/project identities and least privilege.

Approved Tools

Controlled compute/software/network policy.

No Unreviewed Raw Export

Output review/disclosure control before release.

Monitoring & Audit

Session, query, file and output access logged.

MQL Examples

mqlCopy

# Future MQL transformation concept
FIND Observation
  WITHIN cohort "t2d-v1"
  NORMALIZE units USING UCUM
  MAP codes USING LOINC
  VALIDATE provenance, quality
  OUTPUT model OMOP

Conceptual future MQL only; not an active production contract.

Best Practices

Pseudonymize only after approved extraction scope is known

Keep re-identification keys separate and highly restricted

Use reproducible code/config for transformations

Version all outputs

Perform disclosure review before exports

Do not copy production data into unmanaged analyst environments

Developer Integration Surface

Surface	Operation / resource	Use	Status
Transformation jobs API	create / status / version	Research dataset pipeline	Planned/versioned
OMOP/CDISC export	approved model output	Research analysis/regulatory	Planned/evolving
Secure environment API	workspace/job/output controls	Governed analysis	Planned
MQL	future transform/query statements	Healthcare-native orchestration	Future/planned

Next Steps

Freeze target data model

Implement versioned transformation

Apply privacy risk controls

Validate quality/lineage

Publish only to secure environment

Review outputs before release