Research Data Pipeline Example

Scenario: Academic medical center needs to convert EHR data to OMOP CDM for a multi-site research study on diabetes outcomes.

The Challenge

Multiple data sources (Epic, lab systems, pharmacy)
Complex OMOP concept mappings
IRB compliance and de-identification
Ongoing data refresh requirements

Wave Solution

1. Source Profiles:

Epic FHIR Condition resources
Epic FHIR MedicationRequest resources
Quest Diagnostics lab results (custom CSV)

2. Target Profiles:

OMOP CONDITION_OCCURRENCE table
OMOP DRUG_EXPOSURE table
OMOP MEASUREMENT table

3. Complex Mapping Example:


# Generated by Wave for diabetes condition mapping
def map_diabetes_condition_to_omop(fhir_condition: dict) -> dict:
    """Map FHIR Condition to OMOP CONDITION_OCCURRENCE"""
 
    # ICD-10 to OMOP concept mapping
    icd10_to_omop = {
        'E11.9': 201826,  # Type 2 diabetes without complications
        'E11.40': 4193704, # Type 2 diabetes with diabetic neuropathy
        'E11.21': 4193323, # Type 2 diabetes with diabetic nephropathy
    }
 
    condition_occurrence = {
        'person_id': get_person_id(fhir_condition['subject']['reference']),
        'condition_concept_id': map_icd10_to_concept(
            fhir_condition['code']['coding'][0]['code']
        ),
        'condition_start_date': parse_fhir_date(
            fhir_condition['onsetDateTime']
        ),
        'condition_type_concept_id': 32020,  # EHR record
    }
 
    return condition_occurrence

Results

✅ 95% automated concept mapping accuracy
✅ 2.3M patient records transformed successfully
✅ Research ready dataset in 3 days vs 3 months manually

Key Takeaways

OMOP Expertise Built-In

Wave understands OMOP CDM structure and concept relationships, automatically generating mappings that follow OHDSI best practices.

Multi-Source Integration

Combine data from different EHR vendors, lab systems, and external sources into a unified research dataset.

Concept Mapping Automation

ICD-10, SNOMED, LOINC, and other coding systems are automatically mapped to OMOP standard concepts with high accuracy.

Compliance-Ready

Generated transformations include de-identification patterns and audit trails required for research compliance.