ECE Project Group minutes 03-14-2016

1

Welcome

BGO

• Due to technical reasons, meeting was unable to be recorded

• New members David Sperzel and Jean-Marie Rodrigues introduced

2

Causality - Due to/co-occurrent with/caused by

BGO

• Clarify editorial guidelines as to best approach to satisfy user requests for uncertain, unknown or unstated causality

  • Expanded definition of "pattern 2 ("due to")

  • Expanded definition of pattern 1 ("simple" co-occurrence)

  • X disorder after exposure to Y material agent event

• Test chosen approach with sample concepts to insure reproducibility

Discussion centered around definition of due to and pattern 1 (simple co-occurrence). Revised definitions for editorial guide discussed and consensus was that these definitions were mutually compatible.

SSH suggested definition of due to be changed to “This attribute is used in full concept definitions of a concept C’ as subconcept of C, having direct or indirect causation by some X as differentiating criterion”. This definition id further modified to “This attribute specializes a concept’s definition by assigning a direct or indirect causal relationship as a differentiating factor from it’s parent".

BGO illustrated some test concepts for evaluating the definition of patterns 1, 2 and 3. He asked for the members of ECE to complete this exercise offline with results to be discussed at the next meeting. BGO's test posted below:

causality test_BG.xlsx

ECH pointed out that Patterns 1,2 and 3 still leave a gap in which 2 disorders may not be co-occurrent or after one another but may represent some degree of overlap. ECH will provide some examples of concepts for which this may be an issue. 

Examples attached below:

loosely-specified temporal or causal associations.docx

3

"Covert co-occurrences"

ECH

• From Ed:

"In a polyhierarchical graph there are theoretically millions of co-occurrences, so I’ve had to prune things down. The approach I’ve taken is to use the members ‘reason for encounter’ and ‘health issue’ parts of the GP/FP RefSet (as candidate plausible ‘life phase descriptions’) and see where they co-occur elsewhere in the data (testing for subtypes which have two or more distinct GP/FP RefSet members as ancestors), on the assumption that these are also ‘plausible life phase descriptions’ that happen to be co-occurrences (whether they say so or not in their FSN).

By example, this approach tells me that because the data contains classes for ‘uveitits’ ,’eye infection’ and ‘infective uveitis’ (the latter not in the GP/FP set), we can theoretically reframe ‘infective uveitis’ in the language of the ECE guidance as ‘uveitis co-occurrent with eye infection’, or arguably ‘uveitis co-occurrent and due to eye infection’. I’m not saying that classes like this should be renamed/remodelled, just that they could.

I’ve then merged this with a comparable treatment of the association relationships (even with this sampling approach the co-occurrences significantly outnumber the associations), and am at a stage where it would be useful to present my findings to the group. I’m not honestly sure whether the results so far help or cloud the issues, but I do think it provides a new perspective".

ECH gave a presentation on covert co-occurrence in which he argued that due to multiple inheritance many concepts may be interpreted as co-occurrent and some causal even though the FSNs do not clearly indicate this. 

As an example, should infections of an anatomic site be reinterpreted as an infection co-occurrent with inflammation of that site and based on CLPs as including conditions, be modeled with 2 role groups.

Further discussion required.

4

Drug overdoses

YGA

Should drug overdoses be disorders or events?

Due to time constraints, tabled for next meeting

5

Addition of new temporal relationships

BGO

• Addition of during, during AND/OR after, ?before

• Extending range of occurrence (or add new attribute) to cover additional temporal periods including 307157008 |Temporal periods relating to activities (qualifier value)|, 307158003 |Temporal periods relating to physiological functions (qualifier value)| and 307152002 |Temporal periods relating to procedure (qualifier value)|

Due to time constraints, tabled for next meeting

6

New business

BGO

• Meeting frequency: Do we need to move from monthly to biweekly meetings?

• Meeting time: Change in time of meeting with DST?

• Any other topics

Meeting frequency to be increased to q 2 weeks, 1st and 3rd Monday of every month at 1900 UTC. Next meeting will be March 28th.