Snowray Demonstrates Semantic Similarity Based Search Capabilities
The Snowray Terminology Service by B2i Healthcare takes searching clinical terminologies to the next level by integrating AI based search methods
“Inadequate search-and-browse support is very commonly the root cause of using incorrect code in an Electronic Patient Record. ”— Mark Czotter, CTO, B2i Healthcare
LISBON, PORTUGAL, December 15, 2022/ EINPresswire.com / -- Today’s terminology servers offer keyword search functionality, often with advanced configuration options, but the ever growing size of medical terminologies and ontologies poses many challenges to these traditional search methods.
In his recent presentation at the 2022 SNOMED CT Expo, Mark Czotter, CTO of B2i Healthcare discusses how the Snowray Terminology Service was extended with a search mechanism building on machine learning and semantic technologies.
“Inadequate search-and-browse support is very commonly the root cause of using incorrect code in an Electronic Patient Record. Clinical decision support systems relying on such miscoded data provide less accurate predictions and lead to incorrect decisions dangering patient safety.” - said Mark Czotter, CTO of B2i Healthcare.
Boosting search results with semantically relevant matches can improve the quality of lookups, especially for large ontologies like SNOMED CT. Using artificial intelligence, Snowray eases the burden of the growing complexity of terminology maintenance by offering more relevant picks for value set and concept map authoring. These resources can easily be integrated into clinical healthcare systems via standard APIs, like FHIR.
The Snowray Terminology Service explored the use of semantic similarity in SNOMED CT, using dense vectors to extend lexical searches beyond keyword matches in order to provide more accurate and suitable search results. Additional applicability on other terminologies, like ICD-10, ICD-11 and LOINC are being further explored.
Mark’s presentation is available at https://www.youtube.com/watch?v=X1oPg0r7vp8&t=1s.
The AI assisted functions will be integrated into the public Snowray Terminology Service in early 2023.
email us here