Skip to main content

Part of SNOMED CT PaLM Mapping Best Practice

Mapping software

Terminology mapping is a labour-intensive process, requiring both technical knowledge and subject matter expertise to ensure the quality and clinical safety of the output. Terminology mapping software uses automation to reduce effort and provide functionality to support workflow and governance.


Mapping tools

Several specialist tools are available that offer good technical mapping capability. These tools use a variety of techniques to support mapping and can establish the level of equivalence between terms being matched, thereby offering a level of confidence in the validity of the proposed map. Mapping tools also offer flexible functionality to support workflow and governance and can facilitate access to target terminologies and code sets.


Large language models (LLM)

Techniques for source data cleansing section describes how LLMs are valuable in cleansing source data. However, as LLM are probabilistic in nature, during testing, when asked to perform more deterministic tasks where the outcome is fixed (such as mapping local reportables to SNOMED CT PaLM) LLMs were found to be unreliable. They often 'hallucinate' map targets, even when specifically prompted not to. Consequently, LLMs were deemed less suitable for mapping than other methods.


Mapping tool requirements specification

Detailed requirements for building a terminology mapping tool that supports the pathology and laboratory medicine terminology mapping use-case are documented in the SNOMED CT PaLM Mapping Project Requirements Specification. The requirements were developed in parallel to testing a variety of mapping strategies and mapping tools.

The specification is published together with an options paper on SNOMED CT PaLM mapping for NHS England Programmes and Pathology Networks to consider. Authorised staff can email [email protected] to request a copy.


Last edited: 22 May 2025 5:12 pm