This wiki has undergone a migration to Confluence found Here
<meta name="googlebot" content="noindex">

Difference between revisions of "CG Working Group Meeting Minutes"

From HL7Wiki
Jump to navigation Jump to search
m (Clarify timeline for FHIR proposals, minor editorial changes.)
Line 113: Line 113:
 
'''Discussion:'''
 
'''Discussion:'''
 
* The FHIR Management Group has decided, after ballot reconciliation, to proceed to DSTU. That said, they are already looking to the next revisions.
 
* The FHIR Management Group has decided, after ballot reconciliation, to proceed to DSTU. That said, they are already looking to the next revisions.
** New resources need to be done by August. The Quality Assessment review will be done in September. That will allow a ballot to be formed by November for a January ballot for DSTU 2.
+
** If the CG work group wants to propose any new resources, extensions, or profiles, there is a tight schedule in order to make next January's ballot for second DSTU:
*** There is a Wiki page and submission form that needs to be completed to propose any new, core resources.
+
*** Notify the FHIR Management Group of intent in the next month or two.
*** The quality criteria for FHIR encourages at least five examples of each resource.
+
*** Complete new resources/profiles, ''with examples,'' by August.
*** Need to clearly document why a new resource, as opposed to profiling an existing one, is necessary.
+
*** The FHIR Quality Assessment review will be done in September.
 +
*** Ballot will be prepared and submitted by November.
 +
** There is a Wiki page and submission form that needs to be completed to propose any new, core resources.
 +
** The quality criteria for FHIR encourages at least five examples of each resource.
 +
** Need to clearly document why a new resource, as opposed to profiling an existing one, is necessary.
 
** Breadth and depth of DSTU implementations will dictate what FHIR resources go normative.
 
** Breadth and depth of DSTU implementations will dictate what FHIR resources go normative.
** The clinical genomics group has one to two months to prepare any new resources, extensions, and profiles if we want them to be considered in the next DSTU.
+
 
 
* Three levels of FHIR profiles:
 
* Three levels of FHIR profiles:
 
# ''ad hoc:'' developed by one person or group on their own
 
# ''ad hoc:'' developed by one person or group on their own
 
# ''vetted:'' profile submitted to workgroup for modification and subsequently receives HL7 stamp of approval (whatever that means).
 
# ''vetted:'' profile submitted to workgroup for modification and subsequently receives HL7 stamp of approval (whatever that means).
 
# ''balloted:'' profile is taken forward by HL7 to formal ballot for approval. Lays the groundwork for ISO ''et al'' standardization.
 
# ''balloted:'' profile is taken forward by HL7 to formal ballot for approval. Lays the groundwork for ISO ''et al'' standardization.
* The [http://www.hl7.org/implement/standards/fhir/familyhistory-profiles.html | GeneticPedigree Profile], part of the [http://www.hl7.org/implement/standards/fhir/familyhistory.html | Family History Resource] documentation, was developed by Lloyd using the standard V3 Family History model. It has not yet be exercises using sample data, but it's great start.
+
* The [http://www.hl7.org/implement/standards/fhir/familyhistory-profiles.html GeneticPedigree Profile], part of the [http://www.hl7.org/implement/standards/fhir/familyhistory.html Family History FHIR Resource] documentation, was developed by Lloyd using the standard V3 Family History model. It has not yet be exercises using sample data, but it's great start.
 
* The FHIR build process -- taking the source code from GForge and building the documentation -- involves validating the examples against the profiles. Therefore there is high confidence that any example that is provided is well formed and valid.
 
* The FHIR build process -- taking the source code from GForge and building the documentation -- involves validating the examples against the profiles. Therefore there is high confidence that any example that is provided is well formed and valid.
 
** While not in the final documentation, the build process also captures which resource elements were used when validating the examples.
 
** While not in the final documentation, the build process also captures which resource elements were used when validating the examples.
Line 132: Line 136:
 
* The observation resource has no support for genetics at this time.
 
* The observation resource has no support for genetics at this time.
 
* O&O provided persuasive feedback wrt the Observation resource at the San Antonio WGM. As a result, it's expected that it will be modified during the DSTU rev 1 period.
 
* O&O provided persuasive feedback wrt the Observation resource at the San Antonio WGM. As a result, it's expected that it will be modified during the DSTU rev 1 period.
* Distinction between observation and interpretation: an observation documents that someone has a specific genetic sequence (e.g., an allele of a Cytochrome P450 drug metabolizing gene). An interpretation documents expected implications of the observation (e.g., that someone will metabolize Warfarin (aka Coumadin) rapidly).
+
* Distinction between observation and interpretation: an observation documents that someone has a specific genetic sequence (e.g., an allele of a Cytochrome P450 drug metabolizing gene). An interpretation documents expected implications of the observation (e.g., that someone will rapidly metabolize Warfarin (aka Coumadin)).
 
* It is TBD if capturing sequence information, either current or Next Gen, might be an Observation resource, a DiagnosticReport resource, or something new.
 
* It is TBD if capturing sequence information, either current or Next Gen, might be an Observation resource, a DiagnosticReport resource, or something new.
  
 
=====V3 Family History=====
 
=====V3 Family History=====
* Lloyd found it odd that there would ever be an ''approximate'' date of death for the patient. It was explained that when traversing a family tree, and conducting risk assessments for each individual, one whose date of death was not know could be treated as the patient of interest. Such an undertaking is done to determine the most likely carrier for a share family mutation.
+
* Lloyd found it odd that there would ever be an ''approximate'' date of death for the patient. It was explained that when traversing a family tree, and conducting risk assessments for each individual, one whose date of death was not know could be treated as the patient of interest. Such an undertaking is done to determine the most likely carrier for a shared family mutation.
  
 
==2013==
 
==2013==

Revision as of 19:53, 28 January 2014

2014

HL7 WGM San Antonio January 2014

Tuesday, Q1: Introductions, Planning, Preparation

Attendees:

  • Scott Bolte, GE Healthcare
  • Howard Strasberg, Wolters Klewer
  • Bob Milius, National Marrow Donor Program
  • Grant Wood, Intermountain Healthcare
  • Siew Lam, Intermountain Healthcare
  • Clem McDonald, National Library of Medicine

Discussion:

  • Mollie will lead discussion with A/P in Q4.
    • Will include work shes doing on cancer biomarkers and work being done with CAP (College of Anatomical Pathologists)
    • Looking to LOINC for NextGen sequencing codes
  • Marrow donor needs
    • HLM - Histoimmunogenetics Markup Language
    • Need to accommodate existing HML user base conventions
    • Looking to incorporate into CDA narrative parts
    • Might translate or encapsulate into structured data part of CDA

Tuesday, Q2: FHIR Knowledge Sharing

Attendees:

  • Scott Bolte, GE Healthcare
  • Howard Strasberg, Wolters Klewer
  • Bob Milius, National Marrow Donor Program
  • Chuck Swegle, GE Healthcare
  • Phil Pochon, Covance
  • Siew Lam, Intermountain Healthcare
  • Clem McDonald, National Library of Medicine

Discussion:

  • Much discussion about the FHIR tutorials (available at GForge) and the FHIR connectathon
  • Grahame's test server is at http://hl7connect.healthintersections.com.au/open. It is both a FHIR endpoint and a web server. As such, it is a great learning resource to explore the resources, searches, and sample RESTful URLs.
  • FHIR profiles can both constrain and extend a resource.
  • Constraints are implemented using XPath. Can enforce complex criteria such as maternal cousins being invalid when the mother is an only child.
  • Not clear if HL7, or HL7 working groups, will take on the role of endorsing or approving profiles. If so, could end up with multiple tiers of profiles: HL7 core, HL7 approved, 3rd party.
  • Providers, vendors, or even governmental bodies could have their own FHIR servers with their own FHIR profiles that their constituents could use.
  • Clem demonstrated a JavaScript package, developed by the NML (National Library of Medicine), that allows exploration of coding systems. Could complement the pedigree work provided by the Surgeon General's My Family Health Portrait web site.

Tuesday, Q3: GTR, New, merged CG Domain Analysis Model

Attendees:

  • Scott Bolte, GE Healthcare
  • Bob Milius, National Marrow Donor Program
  • Grant Wood, Intermountain Healthcare
  • Siew Lam, Intermountain Healthcare
  • Amnon Shabo, IBM (remote)


Discussion:

  • Free form discussion about the Clinical Genomics Clinical Statement; an evolving document circulated multiple times by Amnon to the CG distribution list over the last month.
  • Discussed how the CG Clinical Statement could/would map to FHIR. Since both use UML style cardinality, external code sets, aggregation & reference, it was thought to be feasible.

Tuesday, Q4: Joint with A/P

Attendees:

  • John David Nolen, Cerner
  • Victor Brodsky, Weill Cornell Medical College
  • Joe Birsa, GE Healthcare
  • Scott Bolte, GE Healthcare
  • Bob Milius, National Marrow Donor Program
  • Grant Wood, Intermountain Healthcare
  • Siew Lam, Intermountain Healthcare
  • Phil Pochon, Covance
  • Amnon Shabo, IBM (remote)
  • Mollie Ullman-Cullere, Dana-Farber Cancer Institute (remote)
  • Jeff Knapp (remote)

Wednesday, Q1: Family History

Attendees:

  • Scott Bolte, GE Healthcare
  • Bob Milius, National Marrow Donor Program
  • Robert Worden,
  • Grant Wood, Intermountain Healthcare
  • Phil Pochon, Covance
  • Mollie Ullman-Cullere, Dana-Farber Cancer Institute (remote)

Discussion:

  • Talked about FH (Family History) and upcoming Q4 meeting with Lloyd.
    • Need to identify sources of sample data, covering diverse clinical situations.
    • CAP, NMDP were both identified as potential sources.
    • Hughes Risk Apps, as the leading commercial adopter of the V3 FH, should also be consulted.
    • The approved | V3 Implementation Guide clearly should be used.
  • Intermountain is working with Cerner on their family history and genomic capabilities.
    • Meetings are slated for February. Not sure if they will be open to external, interested parties.
    • Mark Hoffman, who has worked with the HL7 CG work group in the past, is still involved in Cerner's genomic efforts.

Wednesday, Q2: Joint w/O&O, AP, II

Wednesday, Q3: Project Planning and CG Roadmap

Wednesday, Q4: Joint w/FHIR

Attendees:

  • Scott Bolte, GE Healthcare
  • Bob Milius, National Marrow Donor Program
  • Lloyd McKenzie, HL7 Canada
  • Jeffery Ting, Systems Made Simple
  • Lee Unangst, Academy of Nutrition and Dietetics
  • Kay Howarter, Academy of Nutrition and Dietetics
  • Grant Wood, Intermountain Healthcare
  • Steve Fine, Cerner

Discussion:

  • The FHIR Management Group has decided, after ballot reconciliation, to proceed to DSTU. That said, they are already looking to the next revisions.
    • If the CG work group wants to propose any new resources, extensions, or profiles, there is a tight schedule in order to make next January's ballot for second DSTU:
      • Notify the FHIR Management Group of intent in the next month or two.
      • Complete new resources/profiles, with examples, by August.
      • The FHIR Quality Assessment review will be done in September.
      • Ballot will be prepared and submitted by November.
    • There is a Wiki page and submission form that needs to be completed to propose any new, core resources.
    • The quality criteria for FHIR encourages at least five examples of each resource.
    • Need to clearly document why a new resource, as opposed to profiling an existing one, is necessary.
    • Breadth and depth of DSTU implementations will dictate what FHIR resources go normative.
  • Three levels of FHIR profiles:
  1. ad hoc: developed by one person or group on their own
  2. vetted: profile submitted to workgroup for modification and subsequently receives HL7 stamp of approval (whatever that means).
  3. balloted: profile is taken forward by HL7 to formal ballot for approval. Lays the groundwork for ISO et al standardization.
  • The GeneticPedigree Profile, part of the Family History FHIR Resource documentation, was developed by Lloyd using the standard V3 Family History model. It has not yet be exercises using sample data, but it's great start.
  • The FHIR build process -- taking the source code from GForge and building the documentation -- involves validating the examples against the profiles. Therefore there is high confidence that any example that is provided is well formed and valid.
    • While not in the final documentation, the build process also captures which resource elements were used when validating the examples.
    • The need for V3 style implementation guides, created independently of the V3 structure, should be reduced. The aforementioned validation of examples, and the insistence that at least five examples be provided, implies that more documentation will be available as a profile/resource is developed.
Observations, Interpretations, and Diagnostic Reports
  • The observation resource has no support for genetics at this time.
  • O&O provided persuasive feedback wrt the Observation resource at the San Antonio WGM. As a result, it's expected that it will be modified during the DSTU rev 1 period.
  • Distinction between observation and interpretation: an observation documents that someone has a specific genetic sequence (e.g., an allele of a Cytochrome P450 drug metabolizing gene). An interpretation documents expected implications of the observation (e.g., that someone will rapidly metabolize Warfarin (aka Coumadin)).
  • It is TBD if capturing sequence information, either current or Next Gen, might be an Observation resource, a DiagnosticReport resource, or something new.
V3 Family History
  • Lloyd found it odd that there would ever be an approximate date of death for the patient. It was explained that when traversing a family tree, and conducting risk assessments for each individual, one whose date of death was not know could be treated as the patient of interest. Such an undertaking is done to determine the most likely carrier for a shared family mutation.

2013

Meeting Minutes - HL7 WGM Phoenix January 2013

 Media:HL7_Clinical_Genomics_-_WGM_Phoenix_2013_Minutes.doc

2012

Meeting Minutes - HL7 WGM in Vancouver - May 2012

  • CG Monday Q3 Joint with CIC
    • Attendance: CG - Joyce and Amnon, CIC - Anita and Ed
    • M&M is in the process of coming up with some DAM guidance for HL7. It is our understanding they will be starting with the document CIC created on the subject.
    • Lloyd McKenzie is spear-heading this effort. The DAM guidance will be based on the document produced by CIC on Feb. 2011.
    • Spoke about the different projects with a focus on how the disease projects in HL7 are interfacing with the CDISC therapeutic area standards. The modeling effort in HL7 is focused on clinical and administration domains
    • Example interaction for cardiovascular standards:
    • CDISC ACS (Acute Coronary Syndrome), HL7, FDA working through C-PATH (funded by Gates Foundation) work on identifying disease data end points and DAMs.
    • Subsequently these end-points are presented to the CDISC SDS team for inclusion into the standard.
    • HL7 WGM Vancouver May 2012


  • Tues Q3
    • Attendance: Amnon Shabo, Bob Milius, Jill Kaufman, Joyce Hernandez, Phil Pochon, Grant Wood
    • Jill suggests that the CG WG consider doing a CME tutorial at the next WGM in Baltimore.
    • Amnon reviews the agenda.
    • Joyce suggests we get AP more involved in Omics DAM development.
    • Amnon gives a quick description of the new HL7 initiative called FHIR.
    • A deeper discussion is scheduled for Wed Q2.
    • Amnon also gives a quick description of an outside effort called iPOP. A deeper discussion is scheduled for Thurs Q2.
    • He then went through a review of CG activities.
    • Joyce had a question about O&O Lab 3.0 and genetic variation. Phil shared related efforts with CDISC and RCRIM.
    • Lab V3 no research or genomics – also need to upgrade specimen semantics
    • Research labs could use V2.5.1 message
    • Sponsors might be more interested in a V3 XML solution – add clinical research concepts – study, visits, subject id etc. into O&O Lab V3 message. Unclear what is going to be used on the Research side.
    • Can we use FHIR?


  • Tues Q4
    • Attendance: Amnon Shabo, Bob Milius, Joyce Hernandez, Phil Pochon, Lisa Schick, Grant Wood
    • Amnon continued his slides on CG activities, reviewing CG CMETs and the Pedigree specification.
    • Genetic Variation - Take out specimen handling and identifiers that will be revised in the current Specimen project
    • The Phenotype CEMT model is okay since we are pointing to a separate model
    • Associated Observations are used through controlled vocabulary
    • The Pedigree model was successfully reaffirmed in HL7 balloting, with Release 2 and a US-specific Implementation Guide under development; Exchange of pedigrees in CCD was discussed; Pilots: Hughes Risk Apps and Surgeon General.
    • The V2 Implementation Guides publishing status was given by Grant. Official HL7 publishing forms have been completed, and the TSC will (likely) approve the guides at their next meeting.
    • Amnon reviewed the CG DAMs (Clinical genomics including gene expression, and clinical sequencing). Amnon concluded with a review of CG work on clinical sequencing workflows and use cases - from summary slides written by Mollie.
    • Bob presented his work on HLA tissue typing. This comes from the NMDP matching program, and interest in adopting HL7 standards (CDA GTR) to transmit and receive electronic data with transplant sites. He gave an explanation of the NMDP HLA tissue typing nomenclature version 3. Their Allele types cover up to 6 exons and 800 SNPs. Bob reviewed the NMDP gold and silver standards for reporting data. He showed how he is attempting to put this data into a CDA GTR and Genetic Variation message. They have 150,000 allele codes, but are looking for a different code format. They are looking to use GL Strings for the report, and QR codes that can take you back to the original sequence.


  • Wed Q1
    • Attendance: Amnon Shabo, Joyce Hernandez, Phil Pochon, Lisa Schick, Grant Wood
    • This was a joint meeting with O&O, II (Imaging Integration) and AP (Anatomic Pathology). AP is discussing using the same unique identifier throughout the sample collection, sample prep, lab accession, and result reporting process. They are discussing using a dumb (random number) or smart identifier (meta data, code that describes the specimen). Either way, the container is what is labeled.
    • APSR – Anatomic Pathology Structure Report is a joint IHE-HL7 AP initiative and includes 21 CDA templates (Document Content Modules) and 490 observations and procedure templates
    • The discussion then switched to some of the molecular biology work based on CAP protocols and AJCC (American Joint Committee on Cancer)
    • They are working on a design to implement "value sets" a constrained set of terminology.
    • Based on high-level requirements it seemed they wanted to have a unique identifier that corresponds to the status of the samples as it is processed: bio-specimen collection, resection, and glass slide.
    • Various techniques of associating an id to a specimen were discussed – affixing a barcode on a specimen to bar-coding the container.
    • Two current ideas regarding the specimen ids were mentioned:
      • Use of a pseudo random number
      • Building intelligence in the key. The ramification is that the key would need to appropriately be sized to handle the required built-in intelligence.
      • Specimen process is current documented in DICOM 5.17 specs which are currently informative.
      • DICOM is using OIDS to provide unique identifiers for specimens and containers.


  • Wed Q2
    • Attendance: Amnon Shabo, Joyce Hernandez, Lisa Schick, Grant Wood
    • Amnon reviewed the status of CDA GTR
      • Joyce suggested separating human and viral genomics in the Genetic Variation section.
      • Amnon describes the concept of clinical genetic statements. He suggests we consider doing dynamic binding of answer lists.
      • Amnon then reviewed what he calls the Layout Challenge (see slides in the overview presentation at our page in the HL7 main site), the question of the separation of CDA rendering and logic which CDA R2 core model cannot accommodate but style-sheets could work around this limit.
      • A motion was made to adopt GTR sections, get LOINC codes for those sections, then offer style sheets (CAP, ACMG, JMD); the motion received no objections.
      • Amnon plans to do tissue typing pilot with Bob and two transplant centers.
      • Grant plans to do a pilot with Pathology group at Intermountain, and hopefully Myriad and ARUP.
      • It was mentioned that Don Rule has also talked about doing a CDA GTR pilot on previous CG calls.
    • Family history:
      • FH in Meaningful Use Stage 2 was discussed.
      • Amnon continued slides that reviewed the Pedigree Release 2 Implementation Guide work (*** For more details, see slides in the overview presentation at our page in the HL7 main site).
      • The question of Pedigree and the new HL7 FHIR was discussed.
      • Joyce commented that the IG should show how to move from CCD to full Pedigree.
      • Stage 2 proposed rule does not require pedigree and it talks about family history for the first time.
      • EHR system structured data for the person's first degree relatives mother, father, siblings and children.
      • Meaningful Use Final rule not released and won’t be for months – expectation is that it will be available for the fall time-frame.
      • HITSP recommended use HL7 CCD to exchange data between EHR and PHR and Pedigree to exchange family history information between EHR/PHR systems and decision support applications, and portions of these recommendations are utilized in the new US guide


  • Wed Q3
    • Attendance: Amnon Shabo, Joyce Hernandez, Phil Pochon, Lisa Schick, Grant Wood, Elaine Ayers (NIH)
    • Omics DAM ballot reconciliation and comments were addressed. There was not enough time to complete this process.
    • Since the Analytic phase includes quality control information we need to expand on the data examples to ensure complete coverage.
    • Some QC areas briefly discussed:
      • Hydra well – distilled water hot plant DNA – ensure it is not corrupted.
      • Controls used in fluorescent to detect contamination. Readings are not about the subject (i.e. patient) but quality/purity of material.
      • Mechanical failure
      • need to have a complete quality profile
    • The items above were deemed part of the analytic phase.
    • Post analytic phase would focus on reporting what genes were over and under expressed and tie these readings to a medical statement about the patient.
    • Examples: classify cancer and it phenotypic effect. Correlate this disease/effect to the gene expression profile.
    • One question that came up in the discussion was whether we need a direct association from findings to "updater".


  • Wed Q4
    • Joint meeting with AP.
    • Amnon presented the collaboration with AP to harmonize the GTR and APSR.
    • Discussion centered on the specimen process.
    • The point was made that taking an image of a specimen would result in that being a specimen too in terms of tracking.
    • Requirements would include specimen attributes (e.g. role, collection date/time) and container attributes.
  • Thur Q1
    • Joint with CIC
    • The team provided updates on the various projects around disease models. They have assigned a team member to act as Project Manager or liaison (for external projects).
      • EMS - Jay
      • Cardio - Brian
      • Schizophrenia - Meredith
      • TB - Anita
      • Anesthesiology - Anita
      • Trauma - Jay
      • CG - TBD
    • CIC is focused on disease analysis models – the main exception is emergency services.
    • Abdul-Malik Shakir is the project manager for the Max Tool – a tool developed by City of Hope. It is hoped that this tool will help in exchanging UML models between modeling tools.


2010

HL7 WGM Rio May 2010

 Media:HL7_WGM_Rio_2010_as.doc

HL7 WGM Phoenix Jan 2010

 Media:HL7 CG WGM Phoenix 2010_latest.doc

2009

HL7 WGM Orlando Jan 2009

 Media:HL7_WGM_Orlando_2009_Jan15.doc