Build confidence in multi-source data quality
CrossQuery provides powerful tools to view FHIR data in tabular relational form, to manipulate such data, and to test and improve data quality.
Suppose, for instance, you have patient data stored in three different systems with three different data formats. First you use our FHIR Transform Engine (FTE) to map all the data formats to FHIR bundles. Then you can use CrossQuery to inspect data from the three sources, side by side in the same table, to rapidly uncover any inconsistency between them - such as missing records, or inconsistent data values.
A simple example, using the query tool to compare two different sets of patient data (denoted by the codes A and B) is shown below:
This simple query against the FHIR resource model goes directly to the different data sources (such as relational databases, or FHIR servers), queries them efficiently in their own query languages, and converts the results to a common tabular form, to show them in one table.
Wherever the two data sources agree exactly on all columns in a row of the table, the two rows are merged to an ‘AB’ row. Any difference results in two separate A and B rows. In this case, you can see a discrepancy in the family name of one patent.
As in most tabular views, the tables can be easily manipulated to sort and filter rows, revealing any inconsistencies between the data sources. A simple query language (expressed in terms of the FHIR resource model) enables you to pick out just the linked resources and fields of interest.
This example is highly simplified, but may still be typical of the kind of exploratory work which is needed for a preliminary scoping of data quality problems encountered when preparing for data migration, consolidation or ongoing integration, i.e. using highly filtered queries to return small numbers of records that can be manually inspected, and picking out a few fields in one or a few linked FHIR resources.
For more in-depth information, take a look at the CrossQuery information on our Technology page.
