Skip to main content

Basic Concepts

What is the Helios FHIR Server?

The Helios FHIR Server is a commercial-grade, clinical data solution based on the HL7® FHIR® standard and runs on Apache Cassandra™, a robust, high availability, distributed database. The Helios FHIR Server is delivered and deployed as an OSGi application running inside Apache Karaf™. Karaf is a lightweight, powerful, and enterprise ready modulith runtime.

Motivations for Building the Helios FHIR Server

As we looked at the FHIR Server landscape, and experienced first-hand the challenges of working on large-scale healthcare projects, several patterns emerged from our customers.

  • Existing solutions did a great job of implementing and exploring the FHIR Standard as it emerged and evolved, however, loading up some of these initial FHIR Servers with real-world patient data sets with volume proved to be challenging - they didn’t scale.

  • In an effort to appeal to the widest-possible market, existing solutions supported several different back-end databases at the same time. While it is possible to abstract away a database layer, the engineering tradeoffs that come along with that abstraction sometimes incur severe performance or usability penalties. In our case, we wanted to take full advantage of the many high-performance, horizontally scalable, distributed architecture benefits that Cassandra has to offer.

  • The FHIR Standard is a large open standard with many supported FHIR Resources, Search Parameters, and Operations. In practice, we found that those using the FHIR Standard rarely needed to use it in its entirety. In reality, many applications needed a relatively small subset of FHIR Resources, and of those Resources used, a relatively small set of Search Parameters were truly required to effectively support an application. A modular design was required that only Apache Karaf’s OSGi lightweight container could provide. Only need 12 FHIR Resources and 6 Search Parameters for your application? The Helios FHIR Server will load only those OSGi bundles into the runtime that are required, and also will create only those Cassandra tables and types needed. Find that you need 2 more FHIR Resources, 3 more Search parameters, and one additional Operation? Karaf supports dynamically loading those bundles with zero server downtime.

  • As FHIR has grown, we have only seen strong and increasing demand from developers who wish to ask analytic questions of their FHIR data. Often, the solution developers immediately turn to is to transform FHIR data into a format suitable for analytics, and copy the data into a separate analytics solution. We believe it’s simpler and more cost effective to keep the data in one spot, give developers what they are asking for - a data model for FHIR, and let them use their preferred languages and tools. We call this A FHIR-Native Analytics Approach.