[FGED-discuss] Notes from the morning of the Community Based Standards workshop (Part 2)
stoeckrt at upenn.edu
Wed Feb 25 18:31:57 UTC 2015
This is taken from a google doc that participants are writing notes into as we go along.
Who is the ‘community’ that will use these standards?
How do communities drive standards efforts?
(Creators, disseminators, users - 3 types of communities)
Types of creators - e.g. NIST, NLM (MeSH, Organism Taxonomy) - Models - someone owns, may be editorial board, consortium based like IHTSDO; some create on behalf of a community (like HPO).
Content: Multiple disconnected standards
Proposed harmonization efforts
-In biology there is the OBO Foundry
-In clinical standards: SNOMED CT-LOINC and SNOMED CT-ICD-11
Technical: How to distribute standards?
RDF/OWL (Linked Open Data)
APIs (integration in software).
Lynn Etheredge - Data Standards for a Rapid Learning Health System
• examples of failed and successful standards - need networks/ systems where the standards are part of the work flow. (or likely don’t get used and fail).
• core standards, and several layers of standards - needed to get the info about the 3 B bp of genome data into the EHR. Need certified ‘apps’ with built in standards for specialized data collection.
• missionary work may be needed to get consensus.
Jessica Tenenbaum (creator and user)
want to be standards compliant, what standards do we use? started a group to explore the landscape of standards.
Alexa tried a census of communities, Jessica et al tried a census for the standards 0 what are the standards in genomics? (definition gets mushy - … all the way through domain models, application models. …. from a newbie perspective, difficulty in knowing even that some of them have been deprecated.)
How much rigor needed? To share down the hall, to share out at various levels. Therefore, no right and wrong, but how do know what we need for what purposes?
Mentioned the Sea of Standards paper with Melissa and Susanne Sansone.
Biosharing - already does some of this very well (Susanna and Philippe, e.g.)
Standards concierge? sharing wizard?
Anita de Waard
The Maslow hierarchy of standards - you must meet basic needs first.
Reduce/ Reuse/ Recycle standards.
Design standards with use cases in mind
Start with downstream - for what is it used, and then the upstream (tags/ metadata) --
Don’t just dump in everything. Just the ones that will be used downstream.
What are the metrics for successful metadata? How do you know that a standard actually works? (Adoption, and reuse for example)
Q1. “Shimming” is constantly adopting approaches. What can we do to speed up process of working with data?
• Example of the recent ambitious Precision Medicine vision but how do you realize this? We need essential tools.
• Common Data model between NIH, FDA and PCORI is an example. This is common data model for EHR record data.
There is a need for education in this area.
Eric Neumann’s question: how to integrate?
Three technology communities, three paradigms, three types of metrics!
· Database: # records
· Sem Web: Ontologies, triple stores
· IR: usage, evaluation!
Science cyberinfrastucture lagging behind consumer market here.
Need education, not just technology but way of thinking.
Doug Fridsma: There cannot be a singular solution- “no one ring to rule us all”
Modular approach may be best
Anita: Standards as Boundary Objects:
Weakly structured when not used much, strongly structured with more use.
Question: Do we need to be more organized, a place to document all the standards, …
-Do we need a training program for people who develop standards - e.g. an online course or set of documents
-Do we need to organize all the standards and enable very useful graphical browsing to enable easy finding of existing standards, modules that can be reused e.g. https://myit-2015.itdashboard.gov/treemap
biosharing does some of that… is this workshop going to make a set of recommendations that include something like this?
Charles Bailey: Question: Standards life cycle? think more explicitly about it as we devise standards? We haven’t put a lot of work into defining metadata to describe the standards. Also provides more flexibility to get key parts right now, and revise as we learn more, or as the requirements evolve.
How body of data transitions from one standard into another -- end up with legacy data with standard not changing, or data going dark -- one shot vs ability to change over time -- with the standard life cycle. … to develop a healthier ecosystem. [Alexa - the community would have to be very clearly behind this -- ] you get negative points trying to rescue stuff that was novel 5 years ago.
Anita: She agrees with the formal life cycle approach. In 5 years (maybe) - there are tools to pull out automatically the things the standard was developed to expose. Would be helpful for someone (in the standards group) to monitor the technologies available for addressing use case the standard was developed for: perhaps it can be fulfilled without the standard, and the standard is obsolete before it was even launched?
Chris Chute: Mashup of lifecycle and common data elements - Benefits of a parsimonious lego blocks that can be put together. The Common Data Model is a rigid fixed structure which poses problems. Message: look carefully at identifying the lowest level specification of data that satisfies consistency and … so useful for the next use case.
Olivier Bodenreider: Ontology: Same can be applied to ontology integration. Top down, vs integration. UMLS, less elegant, or less pure, but much more practical. What is good enough for connecting existing standards. Keep in mind the use case for integration.
Melissa: Again refers to the 3 different groups of users (starndards developers, end users of the data, and data integrators) that have to interact with the standards lifecyle. Who participates and how in such a distributed network? How to coordinate feedback from multiple sources and multiple end users? Still fragile and broken community connections. Complex in context of this life cycle.
Anita- some general concepts useful for standards development, akin to software development: add to the lifecycle component for the standard (a formal description of) the versioning of standard, use cases, stakeholders who have a vested interest in these use cases.
More information about the FGED-discuss