[Tufts] Step-by-Step GIS Vocabulary Package Usability Validation

[p-talapova]

Step 1: Define the validation objectives

• What are we testing? (e.g., does the GIS Vocabulary cover all key environmental exposure concepts that I need?)
• What decisions will the results support? (e.g., integrating GIS exposure data into OMOP analytics)

Step 2: Acquire GIS data

• Download GIS dataset (e.g., EJI). Consider complementary GIS sources (e.g., EPA air pollution data).
• Ensure it has a unique spatial identifier, e.g., GEOID (for census tract), lat/long, or zip code.
• Preprocess the dataset, e.g., convert all spatial units (ZIP+4 → Census Tract).

Step 3: Identify patient records matching the GIS data

• extract patient location data from the OMOP location table (person_id, location_id, state, county, zip, latitude, longitude)
• If GEOID (EJI) = location_source_value (OMOP) → Direct Match. If (state, county, zip) in OMOP = (STATEFP, COUNTYFP, GEOID) in EJI → Census Tract Match. If ZIP only available → Crosswalk ZIP → Census Tract using geospatial reference datasets. If Only lat/lon available → Use reverse geocoding.
• Check for address consistency over time (patients may move, the location_history (OMOP CDM v.6.0) table may be needed).

Step 4: Map GIS variables to the OMOP GIS Vocabulary

• Use an interim lookup table to map GIS variables to standard concept IDs (exposure_concept_id) from GIS Ontology
• If applicable, map related units to standard concept IDs (unit_concept_id) from OHDSI Athena
• If a GIS variable has no concept_id, decide whether to add it to the GIS Vocabulary Package.

Step 5: Populate the external_exposure table

• Ensure that each patient in OMOP has a geospatial identifier that can be linked to GIS datasets.
• For each matched patient-location pair assign exposure_concept_id based on the mapped GIS variable.
• Set exposure_start_date (reference date from the GIS dataset)
• Populate value_as_number and unit_concept_id.
• Populate other fields if applicable.

Field Name	Description	Data Example
exposure_occurrence_id	Unique identifier for each exposure record	123456
location_id	Foreign key linking to the location table, indicating where exposure occurred	789
person_id	Foreign key linking to the person table, identifying the individual exposed	100234
cohort_definition_id	(Optional) Links to a defined cohort in research studies	25
exposure_concept_id	Standard OMOP concept_id representing the type of exposure	2052498173 (Percentile Rank Of Annual Mean Days Above PM2.5 Regulatory Standard - 3-Year Average)
exposure_start_date	Date when the exposure event started	2024-01-15
exposure_end_date	Date when the exposure event ended (NULL if ongoing exposure)	NULL (ongoing)
exposure_type_concept_id	Concept ID defining the origin of the exposure record	2052499258 (Government Data)
exposure_relationship_concept_id	Concept ID describing how exposure relates to the person	NULL
exposure_source_concept_id	Source-specific concept ID before standardization to OMOP	90000001
exposure_source_value	Raw exposure value from source data	"EPL_PM"
exposure_relationship_source_value	Raw value describing the exposure-person relationship	NULL
dose_unit_source_value	Source unit before standardization	NULL
quantity	Number of exposure occurrences (if applicable)	1
modifier_source_value	(Optional) Modifier describing the exposure type or intensity	NULL
operator_concept_id	Concept ID defining operator logic (e.g., <, >, =)	NULL
value_as_number	Numerical value of the exposure (e.g., concentration level)	0.8503
value_as_concept_id	Concept ID for categorical exposure values	NULL
unit_concept_id	Concept ID representing the measurement unit	NULL

Step 6: Analyze the impact of GIS exposures on health outcomes.