Considerations for Ontological Alignment between LOINC and SNOMED CT – LOINC extension in SNOMED CT Part 3
Following up on the recent article on what LOINC extension in SNOMED CT and its relevance to the NHS and Unified Test List (UTL), we did some research about the ontological alignment between LOINC and SNOMED CT. We are excited about the ontological alignment since this is a rare opportunity to align two popular international terminologies
Our findings reveal areas to review, extend and potentially restructure key value sets and suggest editorial decisions on model harmonisation for ontological alignment between SNOMED CT and LOINC.
This research is based on two aspects:
- An investigation where we sought to semantically map SNOMED concepts in the UK draft standard for lab medicine, the Unified Test List (UTL) to LOINC concepts and their constituent ‘LOINC parts’.
- Our pre-existing knowledge of SNOMED CT and LOINC and availability of appropriate tooling (Pathnexus) – some of which we have previously written about.
Our findings were far more interesting than just a few bullet points even for a team of informaticians. So we decided that perhaps a journal or conference paper might be a better way to share these findings. So here is a brief summary of what we feel this paper will be about.
Proposed Paper – Ontological Considerations for alignment of LOINC and SNOMED CT
In this paper, we explore the compatibility of the LOINC and SNOMED models for lab test reporting. We examine how easy it would be to undertake the conversion of the full set of modelled concepts in LOINC into SNOMED concepts with their attribute/value logical model in the Observable Entity top level hierarchy. We report our findings and make a characterisation of areas of high congruence and other areas where the models are significantly at variance and editorial decisions on model harmonisation might be required.
We confirm that the underlying ontology informing both LOINC and SNOMED support a high degree of alignment between the two and that metrologic principles are broadly the same. We found that at the parsimonious level of pragmatic modelling adopted by LOINC, the SNOMED model supports reliable inter-terminology mapping but that further editorial efforts will be needed to ensure the models are fully harmonised. We also identify the need to review, refresh, populate and potentially restructure key value sets to support the work. The most crucial areas to review are: methods, specimens/systems, properties and components; specifically (1) underpopulated method value sets, (2) specimen issues such as fluid/body fluid variation and the unresolved distinction between blood, whole blood, serum/plasma, acellular blood, (3) property subsumption, (4) underspecified organism components and components pre-coordinated with for example preconditions, count denominators or properties.
Our early findings also suggest that NHS Digital has followed these precepts well in framing the UTL standard and that it supports reliable mapping and transforms. We have not yet examined ‘non-lab’ LOINC, e.g. vital signs and scored assessment instruments. We note the panel/result headline division and focus on results in the main.
Collaboration on the paper
We do not claim to be experts on both SNOMED CT and LOINC and in fact want to collaborate with experts out there. Here is a sneak peak of just the abstract – and we will share an early draft of the paper with collaborators for review.
Please reach out or drop a comment here if you’d like to collaborate (co-author, review, contribute in other ways etc) on this paper. We have not produced an academic paper in a while – but interestingly some of our research takes us back in time to Manchester University’s work on modularising, normalising and untangling ontologies – Alan Rector and Jeremy Rogers were involved in the OpenGalen project era!
Have you got comments, thoughts on this article or our proposed paper, please drop a comment here.