extracting-clinical-entities

Use OpenMed with a compatible token-classification model to extract biomedical entities from clinical text. The process returns structured named entities complete with labels, character span offsets, and confidence scores for precise analysis.

Yes, you can run named entity recognition on de-identified medical text locally. Run entity extraction after de-identifying clinical documents so sensitive patient data remains secure on hardware under your direct control.

Supported output formats for extracted medical entities include dict, JSON, HTML, and CSV. You can apply confidence filtering, enable sentence detection, group entities, and save spans to JSONL for batch CLI workflows.

OpenMed supports batch processing and JSONL persistence for clinical NLP datasets. You can analyze files from the command line, save extracted entity spans to JSONL, and reuse loaded models for larger document collections.

Clinical entity types extracted from medical notes using token classification include diseases, medications, anatomy, genes, and proteins. It applies named entity recognition to clinical notes, de-identified documents, and research corpora.