Harmonization Engine | Metriport

Metriport raises $26 million to give clinicians insight at the point of care

Thousands of documents in. One record out.

Work with structured data that's clean, standardized, and actionable - no matter the format at its source.

C-CDA · from an HIE …

HL7v2 · ADT feedMSH|^~\&|…PID|1|…^^^MR

PDF · scanned summary

HARMONIZATION

Bundle · FHIR R4

{
  "id": "XXXXX-XXXXXX-XXXXXX-XXXXXX",
  "resourceType": "Bundle",
  "type": "collection",
  "entry": [
    {
      "resource": {
        "resourceType": "Encounter",
        "id": "XXXXX-XXXXXX-XXXXXX-XXXXXX",
        "status": "finished",
        "class": {
          "system": "http://terminology.hl7.org/CodeSystem/v3-ActCode",
          "code": "HH",
          "display": "home health"
        }
      }
    }
  ]
}

C-CDA · from an HIE …

HL7v2 · ADT feedMSH|^~\&|…PID|1|…^^^MR

PDF · scanned summary

HARMONIZATION

Bundle · FHIR R4

Join the teams building on data that arrives usable, whatever shape it left the source in.

Between the source and your query

Arrives as C-CDA · HL0123456789v0123456789 · FHIR · PDF

Harmonization Engine

Four stages between the source and your query.

Source format

Take data in whatever format it exists

C-CDA, HL7v2, scanned documents, and native FHIR are all converted to FHIR R4, with source references retained so nothing becomes untraceable.

Map every code to one vocabulary

Clinical codes map to LOINC, RxNorm, SNOMED, and ICD-10, so the same concept from two sources compares as the same concept.

Resolve one patient across every source

Records that arrive under different identifiers are matched and resolved to the one patient they describe.

Read one record, not fourteen copies of one

Duplicates collapse on consolidation, so the record your team queries is the patient's history told once.

Benefits

Your team reads records, not formats.

The conversion, coding, and dedup all happened before the data reached you.

Coded once, comparable everywhere.

Analytics, quality reporting, and risk work all start from data that is already on one vocabulary, with no per-source mapping project between your team and its first query.

A record, not a pile of near-duplicates.

Networks return the same encounter many times in many shapes. Consolidation collapses the copies, keeps the provenance, and hands your team the history told once.

Your own data enters the same pipeline.

Contributed FHIR resources and uploaded documents run through the same standardization as everything retrieved, so internal and external history read as one record.

How it works

One pipeline, both directions.

What you contribute is cleaned too.

  1. Query the consolidated FHIR record

What comes back has already been converted, coded, resolved, and deduplicated. All four stages ran on retrieval.

  1. Contribute valid FHIR R4 resources

Contributed data must reference the correct Metriport patient; document uploads begin with a patient-linked DocumentReference.

Customers

Chosen by the teams that compared.

A data layer reliable enough to treat on, an evaluation won on completeness, and records that stopped being fragments.

FAQs

  1. What does standardization actually change?

Returned C-CDA documents are converted to FHIR R4, and clinical codes are mapped to LOINC, RxNorm, SNOMED, and ICD-10 so the same concept from two sources compares as the same concept. Source references are retained, so nothing becomes untraceable in the process.

  1. Which FHIR R4 resources and document formats can be contributed?

The standard public interface supports valid FHIR R4 resources and documented clinical-document uploads. Check the current contribution documentation for accepted resources, references, and file formats.

  1. How must references and patient IDs be represented?

Contributed resources must reference the correct Metriport patient and use valid FHIR references. Document uploads begin with a patient-linked DocumentReference.

  1. When is data contribution required?

Data contribution is generally required for exchange customers after relevant encounters. An approved on-behalf-of exception may change that obligation; confirm the applicable production requirements during onboarding.