🔥 Aidbox & Formbox: March 2603 Release

Aidbox 2603 focuses on integrating FHIR workloads with data platforms such as Databricks and BigQuery. The release improves how operational data, analytics, and infrastructure are connected in production environments.

Key changes include real-time streaming of FHIR events to BigQuery, dependency overrides for FHIR packages, and improved support for Databricks Lakebase. The release also introduces the new Aidbox UI by default.

Formbox receives an updated Form Builder with AI agent integration via MCP, email-based form delivery, multi-tenant form sharing, and performance optimizations across the renderer and builder. Multilingual support has been expanded with Arabic and full translation capabilities.

Aidbox FHIR Server

Running Aidbox on Databricks Lakebase

Aidbox can use Databricks identities for Lakebase connections, enabling a more direct integration with Databricks-managed PostgreSQL. This allows Aidbox to use Lakebase as its primary database layer.

Traditionally, FHIR servers run on a dedicated database, while analytics and data platforms are maintained separately. With Lakebase, Aidbox can operate within the same environment as the data platform, reducing the need for duplicated storage and separate data pipelines.

For teams already using Databricks, this simplifies deployment and reduces infrastructure fragmentation. It also aligns FHIR workloads with existing governance, security, and data management practices.

More about Lakebase: https://www.databricks.com/product/lakebase

Real-time analytics with BigQuery

BigQuery is available as an AidboxTopicDestination, allowing FHIR events to be streamed directly into BigQuery using ViewDefinition. This enables a continuous data flow from operational FHIR workloads into an analytical environment.

BigQuery is a columnar database designed for large-scale analytical queries. This removes the need for intermediate ETL layers between FHIR systems and analytics, allowing teams to run near real-time dashboards, reporting, and ML workloads directly on streamed data.

FHIR packages: dependency overrides

Dependency overrides for FHIR NPM packages allow teams to override or exclude transitive dependencies when loading packages. This provides more control over how packages are resolved in real-world environments.

FHIR packages often include indirect dependencies that may conflict or introduce unexpected behavior. Overrides allow teams to stabilize validation and ensure consistent behavior across environments.

For a detailed explanation, see: https://www.health-samurai.io/articles/fhir-package-management#dependency-overrides

Aidbox UI

The new open-source Aidbox UI is now enabled by default and provides a unified interface for interacting with the system. It includes tools for exploring resources, running API requests, querying data, and managing configuration.

The UI includes a Resource Browser, REST Console, SQL Console, FHIR Package Browser, Audit Events viewer, and Settings. It is built with React and TypeScript and is available as an open-source project.

Subscriptions and event delivery

Subscriptions now include additional metadata such as includeEntryAction and entryVersionId. These fields make it easier to distinguish between create and update events in downstream systems.

This improves reliability for integrations that depend on event processing. It also simplifies building pipelines where event type and version tracking are required.

Validation and schema handling

FHIRSchema compilation has been updated to a lazy model, where schemas are generated from StructureDefinitions on read instead of write. This removes the need to store precompiled schemas.

This change reduces data duplication and simplifies storage. It also keeps validation behavior consistent while improving maintainability.

Infrastructure and auditability

Bulk operations such as $import and $load now generate AuditEvents. This provides visibility into large-scale data ingestion processes.

Audit events help track system behavior and support compliance requirements. This is particularly relevant for environments where traceability is required.

Bug fixes and performance improvements

This release includes multiple fixes across GraphQL, validation, terminology, and multi-instance behavior. It also resolves issues with cache replication, concurrent initialization, and artifact registry consistency.

Performance improvements have been applied to CRUD operations and bundle processing. These changes improve system stability under load.

Formbox

Improved Form Builder layout and AI integration via MCP

Formbox introduces an updated Form Builder layout, which is expected to become the default in future releases. In the new UI, widget and form settings are reorganized to make common actions easier to access and reduce configuration complexity. The builder now also supports AI integration via MCP, enabling interaction directly from familiar development environments.

Customization and localization improvements

Formbox extends customization options for the embedded Form Builder, including the ability to hide adaptive form features and select custom renderers via Configuration. It is now also possible to include headers and footers in forms, enabling more flexible structure and branding. Full translation support for the Form Builder interface has been added, including Arabic, which is now also available for multilingual forms.

Delivery and integration

Forms can be sent to patients via email directly from the UI. This enables simple workflows without requiring additional integration layers.

Sandbox mode has been added for testing and development environments. The system also supports configuring multiple terminology servers and customizing loading behavior in embedded forms.

Multi-tenancy and security

Sharing forms across tenants is available from the root level. This simplifies reuse of templates in multi-tenant environments.

Session timeout parameters can now be configured to control automatic logout behavior. Additional configuration options are available for embedded Form Builder instances.

Performance

Form loading time has been optimized for both the Form Renderer and the Form Builder. These improvements reduce latency in embedded and interactive scenarios.

The changes are focused on improving responsiveness in production environments. This is particularly relevant for user-facing workflows.

Explore the full release

Full release notes, documentation, and configuration details are available in the official documentation. These include setup instructions, API references, and examples for each feature.