I am looking for best practices for mapping concepts to another dictionary using the terms: SAME-AS, NARROWER-THAN, & BROADER-THAN.
If I am mapping “Tinea Versicolor” (a fungal infection of the skin) and the dictionary I am mapping to contains:
- “Fungal Infection”
- “Fungal Skin Infection”
- Has anyone developed an internal best practice procedure for mapping an instance like this?
- Should all 3 be mapped?
- Should only the closest term be mapped?
- Does mapping best practices vary based on other factors?
- Does anyone have an suggestions for developing a best practice?
Generally I feel that terminology is not an easy thing for most people to do without additional training and that is why I created the CIEL dictionary to help people. It is MUCH easier to find an interface terminology which is the same thing as the term you are mapping (SAME-AS) and then rely on someone else’s maps to get to the reference terminology (SNOMED, etc.). In your example, it sounds like you are trying to map to a categorical terminology like ICD. You could map it NARROWER-THAN to ALL of those. Fungal skin infection would be the closest but none of them are SAME-AS. However, if you are ICD coding, then there are special rules which dictate which ICD code should be used for a given term, and those might not be totally obvious from the term and the categories.
In general you should only map to the closest map and not any that are in the same hierarchy (so Fungal infection is a parent of fungal skin infection and would be redundant). There are other mapping relationship types which might be added. You should probably not have two SAME-AS maps within the same mapped terminology.
Happy to provide more detail if needed.
Thank you for your response. I will be including your two general suggestions (map to the closet in the hierarchy and map only one SAME-AS) in the mapping guide for our team.
To provide some context: The Ampath Kenya’s local dictionary has many concepts mapped to CIEL concepts, but not all of our concepts having mapping information. In efforts to work with other systems that use CEIL, the Ampath Informatics Form & Concept team be mapping as many local concepts as possible to CIEL. We will approach this from two angles: including mapping information for each new concept created and adding mapping information to existing concepts. Given that this undertaking will be a team effort, I want to standardize our mapping approach.
Do you (or any other community members) know of a group implementing a similar mapping endeavor (local to non-local)?
I so appreciate your time and value your expertise.
This is a common use case, but not many people with the OpenMRS community have tried to do this. Most have simply tried to “Migrate” to the CIEL dictionary so that they can then have access to curated concepts for the rest of their projects. I think that in the private, professional EMR world, IMO (Intelligent Medical Objects) is doing this for nearly 2/3 of the physicians in the US who are using one of the major EHRs (Meditech, Epic, Cerner, eClinicalworks, GE Centricity, Allscripts, etc…)
We have talked about a formal AMPATH migration to CIEL and are also using CIEL as part of the Kenya EMR project. There are several programmatic ways we could do this, in addition to the usual combination of manual mapping with CIEL review. There is also the issue about what to do with concepts that appear in AMPATH but not in CIEL. These should either be created, or if erroneous or no longer used, perhaps retired.
Happy to have a formal discussion about this when I get back into the US next week.