[Roadmap] Geocoding

[jphuong]

1. Development geocoding goals

2. Defining spatial level of deployment (Regional, US-centric, International, Continental)

• US-centric geocoding requirements
• International geocoding requirements

3. Definition of readiness

• Unit testing and build testing procedures
• sample dataset
• benchmarking

4. Vocabulary extension for geocoding attributes

• Documentation and guidance on the metadata strategies
• Attribution of geocoding method (e.g., interpolated or parcel-based methods) and quality metrics

5. Containerized deployments for geocoding

• Postgis
• DeGauss
• Nominatim

[xindyhu]

• The benchmarking test included in the 2024 symposium presentation is very helpful. Do we have plans to expand this to an international setting first before deciding on what to include for the OHDSI GIS WG? For example, I see there is a step to prepare address strings for geocoding, do we know if “street, number, city, postal code” are applicable to other countries? We talked about reaching out to the collaborator from Brazil during the 12/6 meeting

• I want to stress the potential importance of geocoding if it were to be deployed successfully. For example, the All of Us program adopts OMOP CDM but currently address are not geocoded. Making this available would have huge impacts on unlocking environmental health research in All of Us.

• As for maintaining the data catalog, I think it is going to be a huge burden to maintain this as the source data keeps changing. If you only include select numbers of indicators then there will always be someone who asks “how about XYZ”, if you include as many as you can up keeping can be a huge burden and it will also be quite overwhelming for users to access. I guess my suggestion is to leave it up to the individual investigators. OHDSI provides the geocoding tool, what investigators want to do with the lat/long is up to them. They can bring in user-defined layers of environmental exposure or social determinants of health data, and perform a spatial join.

[tdrossi]

I want to echo Cindy's comments, especially the potential importance of geocoding relative to unlocking environmental health regarding exposomics and SDoH. As I’m still learning about the scope of the OHDSI GIS group, I would also like to inquire about a few other topics in this realm:
 

• Regarding Cindy's comment on international standards of “street, number, city, postal code”, can there be an equivalent geolocation measurement used for places and buildings to avoid issues of confusion related to addresses? In particular, creating a digital "polygon" geometry representing the building's boundaries, which can then be correlated to BIM (Building Information Management), can provide accurate measurements regarding IEQ (Indoor Environmental Quality). IEQ includes air quality, lighting, acoustics, thermal comfort, and other environmental factors, not just IAQ (i.e., PM 2.5).
   

• I’m wondering about the ERD of elevation data fields to denote building floors/levels to ensure building location measurements that can be included with data recorded as environmental exposures. This can ensure an option of a delta between exterior and interior environmental measurements on a more granular level (if possible). Would this be a simple geom x,y,z field that is utilized?
   

• If this exposure data is included in an EHR as a Comprehensive Health Record (CHR), what field mapping is planned from an informatics perspective relative to SNOMED, LOINC, etc.? There are some SNOMED and LOINC codes relating to the environment and the built environment. Is there an outline of these fields that this group has created? Agree with Cindy that this type of measurement can be bridged by a spatial join.

• I am also curious if any fields specify whether these measurements are outside or inside a built environment.