Page tree

Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.

SNOMED CT -Exploring pathology challenges

As pathology moves towards an increasingly digitalised platform, and the use of SNOMED CT increases, there are a number of challenges that arise. These individual presentations help to start the conversation relating to these issues specifically. We hope that this will begin a broader conversation relating to the broader pathology implementation using SNOMED CT.

Is this SNOMED? Route finding in UK pathology

The NHS in England developed a cancer registration system in the 1990s. This system accepts data from many sources that include EHRs based in individual provider organisations, laboratory information systems, the Office of National Statistics and organisations providing patient feedback. The current data flow is managed by the National Disease Registration System and the data set related to cancer is known as COSD (the Cancer Outcomes and Services Dataset). These data are used nationally to improve service delivery in cancer care , for research purposes and to contribute to international comparisons of cancer management. The data forms in COSD will be described and the use of SNOMED as one component in the cancer dataset will explored.

A New LOINC – SNOMED Collaboration

In October 2022 SNOMED International and Regenstrief Institute signed a new collaboration agreement for the development of a SNOMED extension which would contain all LOINC content.  This session will describe the goals for the collaboration and provide an update on recent progress, a rough roadmap for development of the extension, technical issues, and logistical and organizational issues.

Prospects of Snomed CT to support workflow in surgical pathology

Surgical pathology is the art to extract information from tissue specimens taken from patients and to present it as reports to guide clinical management and therapy. Surgical pathology workflow involves steps of macroscopic and microscopic analyses supplemented by ancillary immunohistochemical and molecular studies obtaining data on the protein and genetic level, respectively. These steps build on clinical information provided with the specimen (order entry) and on one another.

Synoptic reports have been implemented especially for tumour specimens (by ICCR, CAP and RCPath) to harmonize and structure the information provided. Binding their terminology to Snomed CT greatly supports statistical analyses and cancer registries in their work on a population level.

Currently, specimens and diagnostic information are processed by laboratory technicians and pathologists without substantial professional support by laboratory information systems on the way to diagnosis and report. Data often need to b copied.

Applying Snomed CT in order entry and primary documentation allow to build subsequent steps based on evidence and access decision support directly.

Examples will be discussed how to build abstract rules to support the laboratory workup and the selection of ancillary studies and apply them if the use of Snomed CT is extended from the synoptic report to previous steps of the workflow.

To support workflow, the computation of attribute relationships and postcoordination may play a decisive role.

Using SNOMED for Language Models through Cogstack and MedCAT

Natural Language Processing and Large Language Models have made major strides in past 12 months; as yet most major language modelling are general domain and have not yet incorporated healthcare and biomedical ontology. We demonstrate Cogstack, a healthcare NLP open-source toolkit and community which is designed for real-world use in annotating real world healthcare text allowing large-scale structuring unstructured data to support population health, analytics, improving clinical coding and clinical research. It also includes tools for training and fine-tuning NLP for healthcare and research. It has been deployed in 6 UK NHS Trusts, 1 Netherland UMC and 1 Australian health network, with a number of fast followers adopting it. We will also be able to show the potential for use in AI-based patient simulations.


Widget Connector