BioMedICUS

The BioMedical Information Collection and Understanding System

BioMedICUS is a system for large-scale text analysis and processing of biomedical and clinical reports. The system is being developed by the Natural Language Processing and Information Extraction Program at the University of Minnesota Institute for Health Informatics.

Project Goals

  • Scalability and performance. We use BioMedICUS to process millions of notes here at the University of Minnesota. To do this we need BioMedICUS to have high throughput and to support both machine-level and distributed parallelization.
  • Usability. We try to minimize dependencies and prerequisites that BioMedICUS requires. We release under the permissive Apache 2.0 license and pay close attention to intellectual property issues.

To see what tasks the system supports, look at Components and Outputs. If you are looking for a jumping-in point, see Installation.

Features

RTF Reader

BioMedICUS has an RTF Reader, which has the ability to read and process notes that are encoding in RTF. In addition, BioMedICUS uses RTF formatting information downstream to improve other components.

Acronym Detection

Included in the standard pipeline is an acronym detector, which has the ability to detect and expand acronyms to their equivalent long forms.

Concept Detection

BioMedICUS includes a fast concept detector which labels instances of UMLS Metathesaurus concepts in text.

Downloads

For downloads see the releases page on GitHub. We also make more comprehensive models that require you to have a UMLS license available here.

Contact and Support

For issues or enhancement requests, feel free to submit to the Issues tab on GitHub.

BioMedICUS has a gitter chat for contacting developers with questions, suggestions or feedback.

About Us

BioMedICUS is developed by the University of Minnesota Institute for Health Informatics NLP/IE Group with assistance from the Open Health Natural Language Processing (OHNLP) Consortium.

Acknowledgements

Funding for this work was provided by:

  • 1 R01 LM011364-01 NIH-NLM
  • 1 R01 GM102282-01A1 NIH-NIGMS
  • U54 RR026066-01A2 NIH-NCRR

The following people have made code contributions not represented in the commit history:

  • Robert Bill
  • Arun Kumar
  • Serguei Pakhomov
  • Yan Wang