• 16 May 2023
  • 2 Minutes to read
  • Dark


  • Dark

Article Summary


A past, current, or potential future consumer of health care services.


Broadly used for many purposes.

Detailed Narrative of Logic

  • The following represents a typical sequence of source data integration and patient mastering, ultimately resulting in the population of the Patients natural object. But different implementations may stray from this.
  • Following semantic mapping, any patient referenced in a source data record, having at a minimum a populated Source Local Patient ID field, is loaded into Source Local Patient ID Crosswalk to Data Model Patient ID for master data management.
  • Local patient records are considered to represent the same patient if they share any of the following: (1) social security number; (2) Operational ID (e.g., MRN); (3) combination of Source ID and Source Local Patient Master Identity ID (e.g., a patient identifier that universally identifies a patient within a particular source system); or (4) matching first name, last name, date of birth, and ZIP code. (These are the default matching criteria; they can be customized to include any desired combination of fields.)
  • The contents of Source Local Patient ID Crosswalk to Data Model Patient ID represent the result of this mastering logic: a crosswalk table mapping Source Local Patient ID values to their final Matched Patient ID. The crosswalk -- containing one record per distinct Source ID-Source Local Patient ID pair -- is used to translate Source Local Patient ID values from any source to the mastered Patient ID value. For example, two or more Source Local Patient ID values found to represent the same patient would map to the same Matched Patient ID value.
  • The Local Transform Layer uses the mastered Patient ID, rather than the Source Local Patient ID, to correctly resolve duplicative records and other record collisions that might otherwise be missed if the Source Local Patient ID were to be used, e.g., in data coverage periods.
  • For each source system, a patient object in the Local Transform Layer builds off the mastered data from that source stored in the Source Local Patient ID Crosswalk to Data Model Patient ID object, connecting patient records with other tables containing patient information from that source, if applicable.
  • These Local Transform objects -- one for each data source with any patient information -- is then loaded into Patients, Precursor 1 (All Source Records).
  • Finally, Patients resolves duplicate patient records found across two or more sources, populating patient fields from multiple source records by picking the best values, favoring non-NULL field values and data from more recently updated records. (For example, if the same patient is represented in two data sources, one with date of birth and no address information and the other with address information and no date of birth, the resulting record for that patient will have both date of birth and address fields populated.)

Was this article helpful?