Identify the Natural objects impacted by each source data object

Background and Strategy

The goal of data integration is to create a bridge from the source data to the appropriate downstream Natural objects. Typically, the data from a single source will only represent a fraction of the concepts covered by the full set of Natural objects (see the full list here: Natural Objects). Identifying this subset of Natural objects is essential for developing an appropriate integration plan, starting with semantic mapping.

A reasonable starting point is the heuristic that source systems typically represent one of three types of data:

(1) claims data, i.e., data representing the payor's perspective of the financial transactions associated with the delivery of health care, typically received from a payor;

(2) billing data, i.e., data representing the provider's perspective of the financial transactions associated with the delivery of health care, typically received from a provider's billing system; and

(3) EMR data, i.e., data representing the clinical and operational activities of a provider organization associated with the delivery of health care, typically received from a provider's electronic medical records (EMR) system, but also sometimes from consumer-facing systems.

The contents of claims and billing data packages are often quite predictable. In a typical claims data integration project, for example, the following Natural objects will likely be impacted:

Patients
Providers
Plans
Payors
Institutional Claim Headers
Institutional Claim Service Line Items
Institutional Claim ICD Discharge Diagnoses
Institutional Claim ICD Procedures
Professional Claim Service Line Items
Professional Claim ICD Diagnoses
Pharmacy Claims
Patient-Plan Timelines of Plan Membership
Patient-Source Timelines of Data Coverage

The following Natural objects are also sometimes needed for a claims data package, though less frequently:

Patient Timelines of Time-Varying Features
Program Entities
Patient-Program Entity Affiliation Timelines
Provider-Program Entity Affiliation Timelines

Determining which Natural objects are involved is a matter of combining a good understanding of the source data -- obtained principally from the earlier data exploration and validation tasks -- with a good understanding of the concepts covered by the different Natural objects.

Each Natural object in the Ursa Health Core Data Model has a Semantic Mapping Template containing the superset of fields needed to integrate source data related to that concept. So, one practical benefit of making this determination at this stage is to facilitate the selection of appropriate Semantic Mapping Templates when configuring each Semantic Mapping object.

Ultimately, each source data object containing information needed for analysis should be associated with one or more Natural objects. In the rare cases where there is no existing counterpart in the Natural Object Layer for source data, the creation of a new Natural object should be considered. (If a new Natural object is created, an analogous Semantic Mapping Template for that object should also be created.)

Detailed Implementation Guidance

1. The Semantic Mapping Templates associated with each Natural object are easily identified by their names. For example, the Semantic Mapping Template for the Patients object is named Patient Fields. The full list of Semantic Mapping Templates can be found in the Metadata Manager zone.