KIWI
An LLM-based Clinical Information Extraction System
An advanced medical NLP tool designed to automatically extract information from medical texts.
Try it now
What is Kiwi?
Kiwi is a comprehensive clinical information extraction system using open-source large language models to automatically extract information from medical texts, transforming unstructured text data into structured, analyzable formats to facilitate a deeper understanding and research in medical informatics.
Kiwi is developed and maintained by the Clinical NLP Lab at the Department of Biomedical Informatics and Data Science (BIDS) at the Yale School of Medicine.


Features
Kiwi excels in extracting entities, relationships, and mapped concept IDs from medical texts, using state-of-the-art (SOTA) performance models. The tool is designed for high efficiency, with the capability to process domain-specific texts in less than two seconds.



How does it work?
Powered by the most effective pipeline of three NLP tasks.

