medspacy

Why we included this project

Teams building clinical text pipelines will find medspacy a practical companion to spaCy rather than a replacement for it. It bundles the pieces that generic NLP libraries usually leave out of the clinical domain: sentence segmentation tuned for notes and section detection, plus a ConText implementation that attaches modifiers like negation and uncertainty to extracted entities. The modular design means you can pull in just the preprocessor or the postprocessing framework and keep the rest of your spaCy pipeline intact. Rule-based target matching and UMLS concept extraction via QuickUMLS make it workable without a large labeled training set, which suits research groups and informatics teams that need interpretable, auditable extractions from electronic health records. It is still marked beta and English is the most mature language, so plan for some assembly and testing before relying on it in production.