HL7 FHIR

HL7 FHIR

Background material

There's a lot of information around FHIR, and if you aren't already familiar with HL7, it can be a bit intimidating to find your way around.

• First, here are links to powerpoint presentations that were given in recent HL7 FHIR webinars/tutorials, aimed at three different audiences. These may be shared through the Creative Commons license.

  • FHIR for Executives Part 1
  • FHIR for Executives Part 2
  • FHIR for Architects
  • [FHIR for Developers](FHIR for Developers-baltimore.pptx)

• Here is a meeting report by Bob Milius after attending a FHIR Connectathon in May 2016. A brief description of FHIR is presented, and what Bob learned from attending the Connectathon, including a strategy for developing HLA genotyping report.

  • recorded presentation
  • PDF files
    • May 2016 HL7_WGM_& FHIR_Connectathon - Bob Milius

• The most recent version of the FHIR specification (currently being balloted) can be found here:

  • http://hl7.org/fhir/2016Sep/

• HL7 Clinical Genomics Workgroup

  • The HL7 Clinical Genomics Work Group is developing a Genomics Implementation Guidance document, which describes how to use FHIR resources and profiles for reporting genetics results. In that, there's a section on HLA genotyping with a few examples. It's just a draft, but it should give you an idea of an approach we have been exploring.
    • http://hl7.org/fhir/2016Sep/genomics.html#hla
  • An overview of the FHIR activities of the HL7 Clinical Genomics workgroup can be found here:
    • fhirgenomics.org

• ClinFHIR - a wonderful web tool with a GUI to create resource instances and profiles

  • http://clinfhir.com/
  • brief description on how to use ClinFHIR
  • blog about ClinFHIR by the developer, David Hay

• You can also look at the wiki found at

  • http://wiki.hl7.org/index.php?title=FHIR
    There's a "Help/Getting Started" section there that may be useful.

• Finally, there are a number of blogs out there talking about FHIR. One that I've found useful is this:

  • https://fhirblog.com/getting-started/

NMDP hosted FHIR Server

• A FHIR test server operated by NMDP®/Be The Match® Bioinformatics Research. This server is entirely built using HAPI-FHIR, a 100% open-source Java implementation of the FHIR specification.
  • Web interface
    • http://fhirtest.b12x.org/
  • Base URL for REST (POST, GET, etc)
    • http://fhirtest.b12x.org/baseDstu3

Example data

• This is a set of xml files and snippets of python3 code to POST resource instances to a FHIR server.

  • 00_Patient
    contains a simple example of an instance of a Patient resource. Also included are short examples of python3 code that sends the resource instance to http://fhirtest.b12x.org/baseDstu3/Patient. POSTing this to a FHIR server will return the URI endpoint of the resource instance. For example, the URI that is returned might look like<br > http://fhirtest.b12x.org/baseDstu3/Patient/27/_history/1<br> Keep track of the URI if you want to reuse the same Patient with another resource instance.

  • 01_Specimen
    example of an instance of Specimen resource, that represents a buccal swab from the Patient described above.

  • 02_SimpleTransactionBundle
    

    A single transaction bundle that creates a Patient, a Specimen from that patient, a Sequence from that Specimen, and an Observation about a single HLA-A allele. Here are example resources that were created with this Transaction:
    • http://fhirtest.b12x.org/baseDstu3/Patient/53/_history/1
    • http://fhirtest.b12x.org/baseDstu3/Specimen/54/_history/1
    • http://fhirtest.b12x.org/baseDstu3/Sequence/55/_history/1
    • http://fhirtest.b12x.org/baseDstu3/Observation/56/_history/1

  • todo: A buccal smear collected on a swab (Specimen) is collected from a potential donor (Patient) and is sent to a HLA typing lab (Organization) and requests that HLA typing for HLA-A, HLA-B, and HLA-DBR1 (DiagnosticRequest). The lab does DNA sequencing for each locus (Sequence) and assigns allele names (Observation) and generates genotypes for each locus (Observation). These are aggregated into a single report (DiagnosticReport) and sent back to the requestor.

  • todo: (Use case summary from @npearson, as discussed with @bmilius, @jschneid_nmdp, @mmaiers-nmdp) Via a personally authorized onling proxy (e.g., Unum), an authenticated registry member (what HL7/FHIR calls 'Patient') requests her/his registry-held genomic and ancillary data (latter e.g., contact detail, use consent scope, etc.), in what HL7/FHIR likely calls a 'DiagnosticRequest'. Registry then returns requested data to proxy, who parses and interprets it (latter by first assessing which components of registry member's data such do or, if ambiguous (e.g., serotype versus high-resolution calls), do not suffice to answer each of a set of registry member-configured interpretive questions, and privately returns such interpretation, along with genomic and ancillary dta, to registry member. Registry member can then review interpreted data and ancillary metadata, and edit particular components of ancillary metadata (e.g., contact info), which latter get returned to registry and updated in registry member's record.

A possible FHIR strategy for reporting HLA genotyping

• a DiagnosticReport contains a collection of Observations, each referring to a separate locus
• One locus expanded to show how genotypes, alleles, and sequence are related for a particular locus