Transformers powered REST API for encoding EHRs into SNOMED-CT medical standard terms.
- Encode arbitrarely long EHRs written in natural language into SNOMED-CT formal concept model expressions.
- Identify terms and relations between them, including qualifiers, observable entities, body structures, findings, substances, numeric values and more.
- Achieved 0.91+ F1-Score on novel spanish EHR texts from a variety of medical specialties.
- State of the art NLP techniques based on Transformers like Semantic Textual Similarity and relations extraction.
- Robust encoding of terms regardless of typos, acronyms or abbreviations.
- Fully exposed as REST API.