Researchers have developed Hyperbolic Clinical Ontology Embeddings (HCOE), a novel method for representing clinical concepts that preserves medical code hierarchies. HCOE maps existing BioBERT embeddings into a Poincaré ball, integrating ontology-guided contrastive learning with hierarchical aggregation. This approach has demonstrated superior performance in predicting clinical relations, transferring hierarchical knowledge, and improving outcomes in tasks such as mortality prediction and medication recommendation on the MIMIC-IV dataset. AI
IMPACT This research could lead to more accurate clinical predictions and recommendations by better leveraging hierarchical medical data.
RANK_REASON The cluster contains an academic paper detailing a new method for representing clinical concepts using biomedical language models. [lever_c_demoted from research: ic=1 ai=1.0]
- Anatomical Therapeutic Chemical Classification System
- BioBERT
- Clinical classifications software for identification of codes
- HCOE
- Hyperbolic Clinical Ontology Embeddings
- International Statistical Classification of Diseases and Related Health Problems
- MIMIC-IV
- PheCode
- Poincaré disk model
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
