# Talk to your clinical data: unlocking AWS HealthLake with OpenSharing and Genie One

> Source: <https://www.databricks.com/blog/talk-your-clinical-data-unlocking-aws-healthlake-opensharing-and-genie-one>
> Published: 2026-11-16 18:23:34.994779+00:00

Set up a custom OpenSharing integration to extend AWS HealthLake-managed FHIR data to Databricks, enabling Unity Catalog governance and conversational analytics with Genie One.

Many healthcare and life sciences organizations have standardized on [AWS HealthLake](https://aws.amazon.com/healthlake/) to manage clinical data, but extending that data to other platforms for analytics and AI has traditionally required more effort. For companies seeking to use data living in HealthLake with external analytics tooling, this process has required building export pipelines, duplicating data, and managing governance separately in each location. This has limited the interoperability of HealthLake-managed FHIR data across other tools and platforms.

In this blog, we present a new solution leveraging [OpenSharing](https://opensharing.io/) (A Linux Foundation AI & Data Project) to enable a first-of-its-kind integration between HealthLake and Databricks. This architecture allows users to query HealthLake data live from Databricks, bringing Databricks governance, analytics, and AI capabilities to clinical datasets. By integrating with HealthLake, Databricks can enable organizations to unlock deeper insights from FHIR data and open the door to a broader set of analytics and AI use cases. Furthermore, that same clinical data can now be explored through natural language with Genie One, allowing users to surface insights without specialized technical workflows or data duplication.

The Fast Healthcare Interoperability Resources (FHIR) standard is a data format which is used to exchange health information across different applications and care settings. AWS HealthLake is a managed healthcare data service for storing and working with data in FHIR format. Healthcare organizations often use HealthLake to normalize this FHIR data, as well as for its standards-compliant API for resources such as patient records, observations, and encounters.

HealthLake stores its analytics tables as Apache Iceberg tables in an Amazon S3 bucket owned by the HealthLake service. Customer IAM roles can’t read that bucket directly, so external tools such as Unity Catalog have no direct path to the underlying data. As a result, accessing this data for analytics on external platforms has previously required export pipelines that copy FHIR data out of HealthLake before it can be queried.

While many organizations rely on HealthLake’s purpose-built FHIR data store, extending it with governance, analytics, and natural language exploration has been less straightforward. This new integration with Databricks unlocks valuable new insights from this clinical data.

Our new integration, shown in the architecture below, uses OpenSharing to securely expose data in HealthLake without requiring any export pipelines or data duplication. This allows HealthLake data to be read directly by Databricks and governed by Unity Catalog as a shared catalog.

Today, the only way HealthLake provides direct access to its underlying Iceberg tables is through temporary credentials generated by Lake Formation, and Unity Catalog cannot request these credentials itself. This solution fills that gap with a custom OpenSharing server that runs as a Databricks App, a fully managed serverless deployment. On every table request, the server uses Lake Formation credentials, resolves the table’s current Iceberg snapshot, pre-signs the data files, and returns them in a standard OpenSharing response. Databricks compute then reads the files directly from S3, so no data passes through the server. AWS trusts the app’s own Databricks identity through OIDC federation, protecting sensitive healthcare data. Databricks mounts the share in Unity Catalog like any other share, making the data available for governed analytics, notebooks, dashboards, AI, and natural language exploration with Genie One.

With HealthLake data now readable in Databricks, organizations can bring their clinical data directly under Unity Catalog governance. Unity Catalog becomes the entry point for access, with grants, fine-grained access controls (FGAC), attribute-based access controls (ABAC), and role-based access controls (RBAC) on the shared catalog specifying who is permitted to see and query each table. Access and usage can also be audited like any other data asset in Unity Catalog, making clinical data easier to govern and secure.

That same shared data can now back Genie One in Databricks, broadening analysis of HealthLake data to less-technical users. This means that clinical analysts, operational teams, or strategy leaders can now explore HealthLake data directly, without filing a ticket with the data team or waiting for a custom report.

Organizations can create and curate Genie Agents over the shared catalog by defining core clinical tables, adding plain-language descriptions, and teaching them how to define and interpret clinical concepts like A1C. A user can then ask a question like “How is average A1C trending by month?” and Genie One understands the context, finds the curated Genie Agent, generates the SQL, and returns the summarized answer with a table and chart, all while respecting Unity Catalog permissions. Questions that previously required SQL skills or custom dashboard creation can now be answered in seconds, and the insights living in HealthLake-managed FHIR data become accessible to a broader set of clinical, operational, and business users.

This integration extends HealthLake beyond the analytics paths that have traditionally been available. Organizations no longer need to copy and export data to external analytics tools, or rely solely on Athena or Redshift for analytics. Instead, users now have the option to work with HealthLake data directly in Databricks for governed analytics, notebooks, dashboards, and AI, unlocking new value and insights from clinical data.

The potential is especially significant for Databricks customers managing complex clinical data environments. With this solution, healthcare and life science customers can now apply Databricks-native capabilities on top of their clinical data in HealthLake. With a path to integrate HealthLake and Databricks, these customers can now accelerate at the pace their business needs, without the previous friction of managing complex data pipelines. This opens the door to new use cases for joint HealthLake and Databricks customers. By creating an integration path with Databricks, this solution can help healthcare and life sciences organizations get additional value from their HealthLake data, accelerating analysis and insights across the field.

Talk to your clinical data: get started connecting AWS HealthLake data to Databricks today with our complete reference implementation, an OpenSharing server you deploy as a Declarative Automation Bundle, available as an [official Databricks Industry Solutions repo](https://github.com/databricks-industry-solutions/healthlake-opensharing).

Subscribe to our blog and get the latest posts delivered to your inbox.
