Validate the semantic mapping results

Background and Strategy

Once each semantic mapping object has been run the case-level results should be reviewed.

Primarily, the purpose of this task is to catch easy-to-identify problems related to typos and simple bugs in the Semantic Mapping configuration, rather than deep-rooted issues related to the structure of the source data and how it will be interpreted in subsequent logic.

Key Diagnostics / Heuristics

1. Are the Source ID and Source Data Effective Datetime fields populated for every record?
2. Are critical fields (e.g., Source Local Patient ID) sometimes or always NULL?
3. Is the construction of "Source Local" identifiers consistent across all Semantic Mapping objects associated with the same source?
4. Are external / standard coded values (e.g., revenue codes, ICD-10-CM codes, etc.) formatted correctly?
5. Are source-specific coded values formatted consistently across all Semantic Mapping objects associated with the same source?
6. Do the minimum and maximum values of fields, especially numeric and date columns, fall within the expected range?
7. Do any fields still contain special characters or strings meant to represent missing or NULL data?

Detailed Implementation Guidance

1. Use the Case Review feature -- accessible in the upper right corner of Semantic Mapping object pages -- to review the records in each object.

2. In Case Review, use the Calculate aggregate statistics feature (by clicking the "information" icon found in the header of each column) to generate, for relevant fields, the minimum value, maximum value, and percentage of records with a NULL value.

3. On Semantic Mapping objects whose contents are in any significant way complicated or hard to interpret, create a board named "Useful Case Review Formats" (or similar) to act as a repository for Case Review configurations (filtering, sort order, column order / pinned status) that help to clarify the nature of the object's contents. Pin this board to the left nav and make it visible and editable by all users.