Atlanta Workshop Agenda - 2023-10-21
When
Date: Saturday 21st October, 2023
Time: 09h00 local - 17h00 local (UTC time 13h00 - 21h00)
Goals
Identify feasible and relevant action items for this group in line with the wider SNOMED International strategy for 2020-2025, and beyond.
And primarily continuing:
Consider the issues and challenges faced in meeting the content, quality, and adoption/implementation requirements for SNOMED CT, especially for Terminology Managers and the software they use
Propose possible solutions to solve these challenges
Considering how this may fit into a longer term SNOMED International and the wider community strategy
Where
In Person: Station, https://maps.app.goo.gl/yMzUwPAMoCRmfWb27
Conference call for remote participation:
Join from PC, Mac, Linux, iOS or Android: https://snomed.zoom.us/j/83564054663?pwd=GT5ER0ykauZIr8lwM7DpDkF6Sg1KCa.1
Password: 713332Join by telephone:
Phone numbers - https://snomed.zoom.us/u/kc8g4QYqAJ
Meeting ID - 835 6405 4663
Attendees
Agenda
The agenda for the day. The timings are approximate and topics are expected to move around as the workshop progresses.
Time | Description | Owner | Notes |
|---|---|---|---|
09h00 | Coffee getting, introductions and welcome | @Rory Davidson |
|
09h15 | Background, an update and objective of the workshop | @Rory Davidson |
|
09h30 | 2020-2025 Strategy Update
| @Rory Davidson@Alejandro Lopez Osornio @Kai Kewley | Review of what's been done since Lisbon |
10h30 | Break | ||
11h00 | Terminology Content Syndication | All | A standard approach and process to syndicate/synchronise SNOMED CT content between any terminology server
|
12h30 | Lunch | ||
13h30 | Topic shopping list:
| All | Looking at the differing levels of tooling available through to continuing the push to service based delivery.
Please provide any further topics that you would like to discuss. |
15h00 | Break - Coffee & tea provided | ||
15h30 | Continue disucssions | All |
|
16h45 | Next steps & actions | @Rory Davidson |
|
Comments
Hi Rory,
A second potential topic, if there is any interest by the other NRCs. This is difficult to explain, but it focuses on classification of the terminology. I can best explain it by asking the following questions
What should the "edge" be like between the International edition and a country extension?
How many International concepts should an extension concept subsume prior to it being considered for promotion to the international edition?
Is there a true impact at the implementation of having extension concepts subsume international concepts?
When I first came onboard at the US NRC, there were US extension concepts that were subsuming upwards of 1500 international concepts. This can drastically change classification results at a very high point in the hierarchy thus making it more difficult to identify and resolve classification issues lower in the hierarchy. Over the course of the last 3 years I have tried to decrease that number down to a consistent maximum of 10 International concepts subsumed by a US extension concept. . Once a US extension concept subsumes more than 10 concepts, the depth of the "ragged edge" between the International edition and the extension concepts may become increasing unmanageable. 10 concepts was selected as an arbitrary cutoff point based on the number of concepts in the US extension that subsume less than or equal to 10 international concepts (~1200 extension concepts) as compared to those that subsume more than 10 International concepts (~500 extension).
This is a really interesting topic, I'd not thought of (and you've just sent me down a rabbit hole looking at our data, thanks ).
Looking at the AU release, I see our biggest culprit is "Sexual health concern (finding)" (almost 900 subtypes).
This is an intentionally broad concept, and I suspect would fall foul of the current INT editorial rules.
Extension concepts subsuming Core concepts has been an occasionally contentious topic, but it's unavoidable consequence - the classifier does what it does, with sufficiently defined content.
Personally, I'm not concerned about the number of concepts subsumed, but more about the size of our extension. We'd like to move more of our content into core if possible, and keep the extension thin. Currently promoting content is more trouble than it's worth for us (it actually creates a significant maintenance burden when just the conceptID is promoted). Hopefully this will become less of an issue when we move to managed service.
Very interesting indeed! We also have a lot of concepts in our extension that we have not tried to promote; partly because of the work involved for us, partly because we are not sure how relevant they are internationally, even though most would meet the editorial guidelines; and we do not want to saddle other countries with the burden of translating them But as sharing information across EU countries becomes more common, these Dutch concepts may start to cause problems. Discussion on how to determine whether a concept ought to be promoted; I would certainly welcome that.
There should be no semantic impact to a fully defined extension concept subsuming any number of international concepts.
Hi Rory,
If there is any interest by the other NRCs, could we possible do a deeper dive into release validation, including by not limited to DROOLS and RVF, that takes place as part of the authoring and release processes as it pertains to managed service customers.
Part of the discussion may include:
Current managed service release validation process
How much validation is done in SQL versus using DROOLS and RVF
Process for adding NRC specific DROOLS and/or RVF rules
Potential for e-learning about creating DROOLS and/or RVF rules
DROOLS or RVF shared repository for NRCs
General discussion about QA practices performed by NRCs during release cycle. (would like to know what other NRCs are doing and in what environment.)
Currently, the US is using a combination of Reporting platform, custom ECL scripts, DROOLS, RVF, and SQL server to perform validation checks during the release cycle and would like to determine if there is a way to streamline the process as well as increase the confidence level.