JA EN
shiichan

AWS HealthLake can now auto-link duplicate patient records!

AWS What's New ★★★ View Markdown
#aws#infrastructure

Hi everyone, it's Shiichan! Today I found some news that healthcare data folks will want to know about!

AWS What's New aws.amazon.com

What was announced?

Over on AWS News (AWS What's New), AWS HealthLake announced a new feature called "resource matching," now available in preview!

This feature automatically finds and links duplicate records within a datastore. It works across seven FHIR (Fast Healthcare Interoperability Resources) resource types:

  • Patient
  • Practitioner
  • Organization
  • Location
  • Device
  • RelatedPerson
  • PractitionerRole

The story so far

Duplicate patient records are one of the costliest problems in healthcare data management. A patient's information gets scattered across records, leading to redundant tests, missed diagnoses, billing errors, manual reconciliation, and analytics that mistakenly count one patient as many.

Until now, deduplicating (matching) these records required specialized master data management (MDM) tooling. The point of resource matching is that you no longer need that.

What changes

Once you enable resource matching, every record that's created or updated is automatically evaluated, and matches get connected through FHIR Linkage resources.

The matching logic relies on high-confidence identifiers like Social Security numbers, medical record numbers, and national provider numbers (NPI). It takes into account each identifier's real-world scope and filters out placeholder values, so it avoids false matches.

Your original records are never modified or deleted, so provenance stays intact. Linkages are automatically re-evaluated as source records change, and removed if records no longer match. There's no configuration, matching rules, or third-party tools to set up.

That means you can build accurate longitudinal patient records and trustworthy population health datasets without specialized MDM tooling.

Dive Deep

Resource matching is currently in gated preview, available in these regions:

  • US East (N. Virginia)
  • US East (Ohio)
  • US West (Oregon)
  • Asia Pacific (Mumbai)
  • Europe (London)
  • Europe (Ireland)
  • Asia Pacific Southeast (Sydney)
  • Canada (Central)

You'll need to request allowlist access to use it. Full usage details are in the AWS HealthLake developer guide.

Wrap-up

  • AWS HealthLake added a new "resource matching" preview feature that automatically finds and links duplicate records
  • Supports seven FHIR resource types including Patient, Practitioner, and Organization
  • Matches on identifiers like SSN, medical record numbers, and NPI, filtering out placeholder values
  • Original records stay untouched, with links managed automatically while preserving provenance
  • In gated preview across 8 regions; you need to request allowlist access

If you work on healthcare data infrastructure or have been wrestling with FHIR record deduplication, this update is worth a look!

Related posts