[Use Case]: common data model extension tables for COST and PRICE to assess spatial autocorrelation

[jaanli]

Description

The Common Data Model Workgroup (ohdsi.org/workgroups) has met several times over the past two years to discuss how best to include COST and PRICE tables to enable use cases in observational health research, health equity research, and health care delivery.

Some use cases for these tables (that will likely be a common data model extension) are documented here:

https://github.com/OHDSI/CommonDataModel/tree/payless_health

Infrastructure - Can this be delivered with existing Infrastructure?

No. We need to use protomaps.com because we cannot afford the licensing fees for Google Maps or landsat data.

(Specifically, to pre-train the transformer models such as this geospatial foundation model -- https://github.com/Clay-foundation/model -- we need scalable infrastructure that is open source like @protomaps)

Timeline - What is your desired timeline for completing this work?

We will be completing the initial release this year, in order to assess initial health economics & outcomes research deliverables. Last year we worked with the government accountability office to understand wage depression (https://www.healthaffairs.org/doi/10.1377/hlthaff.2022.01184) and anomaly detection in revenue cycle management (https://www.nber.org/system/files/working_papers/w30946/w30946.pdf).

Credit - How would you and your team like to receive credit for this work?

Yes! Citing our Apache 2.0 licensed codebase would be great (@onefact), and collaborating on peer-reviewed research is usually the best way to contribute.

Support - What kind of support does your team need from the workgroup?

We need help vetting whether we have identified the highest-leverage use cases for assessing geospatial correlation in the COST and PRICE tables related to demographics information and real estate prices.

Examples of work we've done recently in this direction:

https://onefact.github.io/american-community-survey/new-york-area/income-by-race

https://onefact.github.io/new-york-real-estate/

https://onefact.github.io/healthcare-data/

We need help using tools like these methods:

https://connordonegan.github.io/geostan/articles/spatial-me-models.html

With the foundation models for geospatial data that we use from @Clay-foundation, e.g. https://github.com/Clay-foundation/model

Datasets of Interest

Dataset from last year:

https://www.ohdsi.org/2023showcase-410/

Depends On

N/A