Construct a Universal Denial Data Standard

[MikeG112]

Due to both historical timeline of the work on the claims_denial investigation, as well as the underlying methodologies used to report the raw data, the notebooks supporting this investigation are currently a mess

Some things can be fixed without modifying the overall modular approach to the datasets (general refactoring, removing duplicate code, cleaning up documentation and redundancy, etc), but others are issues inherent to lack of standardization of the data.

I propose we move forward by:

• Designing a standardized schema for reporting claims denial data.

  A natural repository to house a community-informed schema development would be this one. The schema should be detailed enough to support capturing much of the rich and diverse metadata that is proposed in some of the existing open source datasets, but be flexible enough to allow using it even for those datasets which report claims data minimally.

• Writing adapter modules for each publicly accessible claims denial dataset, capable of parsing the raw data and dumping a view of it that fits the standard schema.

  Such adapter logic currently exists in the notebook for each dataset, and can probably be repurposed, but they adapt to varying schemas (due to historical convenience/laziness). We can also host copies of the standardized views so everyone need not run adapters to analyze the data.

• Write a single analysis notebook or configurable script, that takes as input a local or hosted copy of a claims denial dataset in the standard schema.