OHDSI GIS Version 1.0
With the intention of splitting up this large project into solvable pieces, Version 1.0 will tackle a subset of the the goals of the GIS working group, which are a documented on the project goals page. The content of the exclusions, listed below, will potentially be included in later versions.
- It is assumed that any external data source (ACS, EPA, NOAA, USGS, etc) has already been transformed into the format of the data model. There has been extensive discussion as to standing up a (PostGIS) staging database for geographically referenced data, pre-loaded with relevant data and functions, but that will not be included in Version 1.0.
- Geocoding your patient address data is a prerequisite process which is not automated by this version. This may, as the issue before it, become a function integrated into a staging database.
- We will not persist records of individual patient level exposures. Although it seems likely we will need this ability for integration with other packages (e.g. patient level prediction), for Version 1.0 we will calculate the relationships from person to geographic attribute data on the fly. In other words, the variable 'Ratio of income to poverty level' will persist as an attribute of the census tract, not the people who live in that tract.
- All geospatial functionality will be performed in R. To simplify the complications of finding a middle ground between universal DB support and the advantages of specific databases (e.g. PostGIS), for Version 1.0 we will stick a simpler approach of using shapefiles and R to persist the geospatial data in a form that can be stored in all OHDSI supported database flavors.
- Spatial calculations performed in R leveraging spatial data classes from 'sp' package
- Polygons stored locally as shapefiles
- All other data, including references to shapefiles, stored in database
