Technology

Talk to FHIR data, if you build the bridge

Databricks claims live HealthLake queries with no pipelines, but you still run the credential bridge.

StartupHub.ai Staff

By StartupHub.ai Staff, reviewed by Daniel Singer, Editor

2 min read

Clinical FHIR tables shared from AWS HealthLake into Databricks Unity Catalog
Credit: Databricks

Databricks says its new OpenSharing bridge lets Unity Catalog read AWS HealthLake FHIR tables live with no export pipeline. The systems in scope are HealthLake Iceberg tables in a HealthLake owned S3 bucket and Databricks compute under Unity Catalog. AWS describes HealthLake as a service to store, transform, transact and analyze health data in minutes, per its product page, which explains why hospitals parked data there.

The obvious read is that export pipelines are dead.

They are not. HealthLake still blocks direct reads to that bucket from customer IAM roles, and Unity Catalog cannot request Lake Formation temporary credentials itself, per the Databricks authors Michael McCarthy, Lily Goldstein and Douglas Moore. The fix is a custom OpenSharing server you deploy as a Databricks App that takes each table request, fetches Lake Formation credentials, resolves the current Iceberg snapshot and returns presigned S3 files for Databricks to read directly. No data passes through the server, and AWS trusts the app through OIDC federation, but you still own that hop, its monitoring and its clinical access path. For the simple story to hold, that credential broker would need to be native, managed and audited without extra code, plus benchmarks on latency and concurrent clinical queries that are still undisclosed.

To be fair, the governance payoff is real when the plumbing works. Databricks says the HealthLake share mounts like any other share, with grants plus fine grained, attribute based and role based controls and full audit in Unity Catalog. The same tables can back a curated Genie Agent where teams name core tables, add plain language descriptions and define concepts like A1C, so a user can ask "How is average A1C trending by month?" and get SQL, a table and a chart under Unity Catalog permissions. That reference implementation ships as a Declarative Automation Bundle in an official Databricks Industry Solutions repo, so pilots can start fast.

I rate the pattern promising and the product claim premature. If you run HealthLake and Databricks, test the App under your own audit and load before you promise self serve A1C answers to the business.

Sources

Get the free daily brief every morning

Every morning the day's AI news, new AI startups and new funding rounds. Free and takes a few minutes to read, unsubscribe at any time.

Join 64,000 other subscribers.

TopicsDatabricksAWS HealthLakeOpenSharingUnity CatalogGenie OneApache IcebergLake Formation

StartupHub.ai StaffDaniel SingerWritten by StartupHub.ai Staff, Editorial teamReviewed by Daniel Singer, EditorAbout

StartupHub.ai Staff, Editorial team

The staff writers of StartupHub.ai, ranging from investment analysts to avid AI tool users, early adopters and critical enthusiasts. Backgrounds span engineering, business and the arts. We hold every piece to rigorous standards of research and review.

Daniel Singer, Editor

Daniel Singer is an AI expert who has been writing about, covering and building AI technology since 2018.

More from Daniel Singer

Test your Agent Readiness now

Paste your website. How well ChatGPT, Claude and the other agents can read and use a site, in about a minute.

Watch

The videos behind the news, one flip at a time

The feed mixes the videos from our articles with startup demos and you flip through them like TV channels for free with no signup.