A large fraction of information in electronic medical records is “locked” in narrative documents, such as provider notes, radiology reports, etc. Natural language processing technology can be used to extract information from narrative documents. However, it remains underutilized, because in many cases natural language processing solutions require advanced computer science expertise and/or expensive commercial software.
Canary (http://canary.bwh.harvard.edu) is a free/open-source software designed to solve this problem. Canary is a GUI-based platform that allows clinicians, researchers and analysts without computer science or software engineering background to develop their own natural language processing solutions. Canary has been downloaded in dozens of institutions around the world and has been successfully used in a number of projects. We will describe Canary and illustrate how it can be used to obtain information from narrative documents you always wanted, but could not reach.
After participating in this activity, the learner should be better able to:
- Understand challenges in developing natural language processing tools
- Create natural language processing solutions using Canary platform
Alexander Turchin, MD, MS, FACMI
Director of Informatics Research
Division of Endocrinology
Brigham and Women’s Hospital
Associate Professor of Medicine
Harvard Medical School
Dr. Turchin is Director of Informatics Research at the Division of Endocrinology at Brigham and Women's Hospital, Associate Professor of Medicine at Harvard Medical School and Director of Clinical Informatics at Baim Institute for Clinical Research. Dr. Turchin is a graduate of Johns Hopkins University School of Medicine and Massachusetts Institute of Technology (Medical Informatics). His research focuses on analysis of electronic medical record data; he uses advanced informatics technologies including natural language processing to study quality of care and outcomes in chronic endocrine diseases. Dr. Turchin is a Fellow of the American College of Medical Informatics and has published over 80 papers and book chapters; his research has been funded by AHRQ, NIH and private foundations.