Using SNOMED in Stardog
Get the latest in your inbox
SNOMED CT is a comprehensive biomedical ontology that describes critical relationships for managing clinical operations in multiple languages. In their own words, SNOMED describes the topics in the ontology as follows:
“SNOMED CT it is not just a coding system of diagnosis. It also covers other types of clinical findings like signs and symptoms. It includes tens of thousands of surgical, therapeutic and diagnostic procedures. It includes observables (for example heart rate), and also includes concepts representing body structures, organisms, substances, pharmaceutical products, physical objects, physical forces, specimens and many other types of information that may need to be recorded in or around the health record.” — SNOMED
Stardog’s platform supports state-of-the-art reasoning capabilities specifically optimized for large, complex ontologies such as SNOMED.
SNOMED models critical relationships within clinical operations; these models can be used directly within Stardog to accelerate Knowledge Graph development.
Stardog’s best-in-class Inference Engine has the most advanced capabilities on the market for processing complex ontologies. Stardog performs reasoning using a combination of Stardog-specific and open-source reasoners Blackout, Pellet and ELK in order to provide the best reasoning performance depending on the expressivity of the input ontology. These reasoners come with optimizations that handle different profiles of OWL 2 and different reasoning functions. Stardog will automatically choose the optimal reasoner that is suitable for the input ontology and the reasoning query being answered.
If SNOMED ontology is loaded into Stardog, ELK reasoner will be chosen automatically. ELK uses a highly optimized consequence-based reasoning algorithm that can also take advantage of multi-core CPUs. ELK can classify SNOMED in 5 seconds on a laptop. Loading SNOMED into ELK from Stardog happens behind-the-scenes and takes another 5-10 seconds. These operations are done only once during reasoner initialization; classification results are cached so subsequent reasoning queries are answered using cached inferences. Stardog’s automated nightly performance benchmarks include SNOMED and other standard reasoning benchmarks to avoid regressions.
Stardog is also highly explainable. By using properties of our query rewriting reasoning algorithm, Stardog is able to present explanations of inferences in a tree-like form so the user can imagine a proof of the conclusion from the premises. The expressivity allows knowledge engineers describe complex domains, such as medicine, in which multiple facts, axioms, and rules interact with each other to infer new facts. These new facts can appear in query results and graph validation reports. In both cases Stardog can explain them using proof trees.
Stardog’s engineering team has extensive experience in ontological reasoning and has contributed to the open-source reasoning libraries mentioned above, which are now integrated into Stardog. Stardog VP of R&D Pavel Klinov is one of the main developers of the ELK reasoner and his PhD thesis from University of Manchester was on probabilistic Description Logic reasoning. Stardog CTO Evren Sirin is the main developer of the Pellet reasoner and his PhD thesis from University of Maryland focused on expressive Description Logic reasoning.
Stardog also has strong partnerships to aid in extracting and tagging valuable scientific data based on SNOMED and other ontologies. SciBite offers a suite of semantic analytics technology solutions which allows complex relationships between entities and their synonyms to be mapped via ontologies, enabling the correct meaning to be applied when these entities are identified within text. SciBite also empowers users to collaboratively build and maintain multiple complex ontologies, including SNOMED, in a single unified tool. Through our partnership, users can access a powerful Knowledge Graph, one that is supplied with data from virtualized silos and external sources and enriched with meaning from various ontologies and best-in-class reasoning. These Knowledge Graphs offer a robust data search and exploration solution necessary for drug research, development, and clinical management.
For more information on SNOMED, check out their site for details.
