| 1 | April business meeting session for this group |
| - Agreement that a face-to-face session in April would be beneficial
- Specify a number of topics to discuss in person from the outputs of the virtual meetings
| Karim to follow up with Jim Campbell |
| 2 | How we apply the outputs from this group |
| - Follow SI governance process regarding inconsistencies in lab extension and specialism content and recommendations from the group
The group can recommend new editorial guidance to SI to meet user needs as long as they do not conflict with editorial policies This group should be working towards modelling lab observable entity content the same way
| Daniel to give a short presentation on time aspect - needs further discussion Create issues log and actions log |
| 3 | Ratio of a substance in two different specimens Ratio of two different measurement properties |
| How do we assess effort vs impact? The requirement to model them accurately for machine reasoning - auto-classifications; erroneous subsumptions Examples: Two different specimens LOINC 73913-6 - Glycine in CSF/Glycine in plasma UK SNOMED CT code 1812741000000102 | Substance concentration ratio of glycine in cerebrospinal fluid to glycine in plasma (observable entity) | Mass/Mol ratio 40486-3 Protein/Creatinine [Ratio] in 24 hour Urine Unit - g/mmol
Brief discussion on options and concept model limitations: - Maintain a small number of concepts that can stay primitive
- Cannot do relative role groupings
- Renaming the attributes / new attributes e.g. 'relative_to_inheres_in' 'relative_to_component'
- Using 'relative concentration property '- problems with subsumption under measurements of the substances - unless placed under ratio observable intermediate primitive
- Model nominator and denominator values as observable entity measurement concepts
We need to figure out the underlying ontological model that represents the dynamic for both use-cases - different specimens; different measured properties LOINC specifies the numerator and denominator in the description but no good way to model it for computability
| Karim to review the modelling options with Daniel and bring back to the group |