Zus — Clinical-Data Aggregation

1What Zus is

Zus is a health-data aggregation platform — “connective tissue” that partners with the national data networks, pulls together a patient’s scattered clinical history, cleans and normalizes it to FHIR, summarizes it, and hands it back to a care team or a developer as a single coherent record. The unit of the platform is the Zus Aggregated Profile (ZAP): an up-to-date, curated, longitudinal record of one patient’s health history, assembled from many external sources. Where Canvas is where care is documented, Zus is where the rest of the patient’s world is assembled.

2Company facts

3What Zus aggregates — the data networks

On top of aggregation, Zus does four kinds of “make it usable” work: FHIR normalization (everything to FHIR R4), data enrichment (consistent terminology sets), intelligent summarization (the Zus Lens — collapses dozens of duplicate documents into what matters), and front-end visualization. Critically, Zus guarantees customers access to all the raw underlying data plus detailed provenance — so the curation never costs you the raw record.

4Architecture — the ZAP, Lens, and UPID

• ZAP (Zus Aggregated Profile). The curated, longitudinal, per-patient record — first-party data, third-party network data, and Lens summaries. Stored FHIR-native (R4), never overwritten.
• Lens. Zus-curated summary layer (condition / medication / encounter / transition-of-care lenses). In the data, Lens tables are distinct from raw resources (prefixed LENS_).
• UPID (Universal Patient Identifier). Identity-resolution service linking duplicate Patient records across sources — the right key for patient-level analytics.
• Provenance / DATA_SOURCE. Every base resource carries where it came from (commonwell, carequality, surescripts, bamboohealth, quest…) — enabling source-aware analytics.

5★ How a Zus client accesses / exports the raw payload

Zus offers five integration pathways, spanning real-time reads to full bulk export — richer than a single “export” button. The right path depends on whether you want to query data Zus hosts or pull the data into your own warehouse.

The bulk / raw path: Data Marts

For the “get the raw payload into our environment” use case, Data Marts are the mechanism. Zus transforms the deeply-nested FHIR JSON into a relational schema (~200 views) optimized for SQL — three table families: Base FHIR tables (relational representations of the raw resources, first- and third-party), Lens tables (LENS_*, the curated summaries), and Data-type tables (reusable FHIR sub-structures — HUMAN_NAME, DOSAGE, IDENTIFIER…). Every base table carries DATA_SOURCE and (except Location/Medication/Practitioner/Organization) UPID.

Data-mart access options

Query data hosted by Zus: Snowflake (private listing or secure share), Snowflake Zus-hosted reader account, Databricks (via the Snowflake/Databricks connector), BigQuery (coming soon).

Export data to an external location: bulk export to cloud storage (AWS S3 / GCS / Azure), external ETL tools (Airbyte, Fivetran via machine users), import into Tuva (via the Tuva ↔ Zus connector).

Freshness

Data marts refresh continuously, all tables ≥ once daily; most reflect changes within 24 hours, with key parent tables (Condition, Encounter, Patient, Practitioner, Organization, Location, Device, DocumentReference, transition-of-care) on an accelerated ~3-hour target. Referential integrity is not guaranteed at every instant (rolling updates) — for strict freshness/integrity, use the FHIR APIs or Zushooks instead.

6Risk-adjustment angle

Zus ships dedicated Risk Adjustment data marts + FHIR models, plus a payor risk-gap CSV import — HCC/RAF-shaped analytics sit natively on the Zus substrate. Directly adjacent to FHG’s reimbursement and clinical-network work.

7The Zus ↔ Canvas relationship

Zus and Canvas have a strategic product partnership, and Zus’s pre-built “Zus for Canvas” integration makes Zus one of Canvas’s named EHR integrations. In a stack running both, the patient’s outside history is aggregated by Zus and surfaced alongside the Canvas record of in-house care. See the Canvas overview →

8Open questions

• Which specific networks/sources are live for a given client’s population, and the realized fill rate per source.
• Data-mart access shape FHG would use (Zus-hosted reader account vs customer Snowflake vs flat-file export).
• How Lens summaries should (or shouldn’t) flow into the FHG spine vs raw base resources.
• TEFCA timing and what it changes about coverage.

9Sources

• Zus Developer Docs — Introduction to Zus
• Zus Developer Docs — FHIR Relational Data Marts
• Zus Developer Docs — Data Mart Access Options
• Zus Developer Docs — FHIR API
• Zus Health — API & Components
• Canvas Medical and Zus — Strategic Product Partnership

Captured 2026-06-09. Re-validated quarterly against docs.zushealth.com. Cross-link target: the FHG Academy data-source track.