FUME docs home
FUME is essentially a data transformation API factory with native FHIR support.
It provides a mapping language and a runtime model for turning source data into FHIR resources and other structured outputs. Mappings combine JSONata-based expressions (extended with FHIR helper functions) for selection, querying, computation, resource resolution and orchestration, with FLASH when the target needs conformant FHIR output.
Any mapping can be called using the RESTful API.
The Playground is the fastest way to try FUME before installing anything. Use the sections below when you need setup guidance, runtime settings, server integration details, language guidance, function lookup, or input-format behavior.
Choose your starting point
| If you need to... | Start here | Why this is the right first stop |
|---|---|---|
| Install or run FUME locally | Docker setup | Community usually starts with Docker, and Enterprise usually starts with Docker as well. On Windows, Enterprise and Designer can use the EXE package instead. If you plan to run Enterprise on Kubernetes, use the FUME Helm chart. |
| Configure runtime behavior | Runtime settings | Start with the product inventory, then drill into grouped setting pages for exact defaults, prerequisites, and deployment notes. |
| Integrate the runtime over HTTP | Server API | Use the workflow guide and OpenAPI pages to choose the right endpoint surface. |
| Learn how to write mappings | FUME language | Start with the shared authoring model, then move into FLASH for FHIR output. |
| Look up built-in helper behavior | Functions | Browse by task-oriented category, then open the canonical function page for syntax, examples, and execution notes. |
| Understand non-JSON request normalization | Input formats | Use the overview and format-specific pages to see how CSV, XML, and HL7 v2 payloads are normalized before evaluation. |