Natural Language Processing to Identify Persons with Carotid Stenosis

The specific aim of this project is to determine the accuracy of a natural language processing (NLP) application at identifying patients with internal carotid stenosis when compared against the gold standard of previously conducted chart reviews. The incidence of stroke is 795,000 per year in the United States. Carotid endarterectomy (CEA) is the only surgical intervention proven in randomized-controlled trials (RCTs) to reduce the risk of stroke. There are 100,000 CEAs performed 1n the United States each year. Within VHA, there are about 4,000 CEAs and an unknown additional number of carotid stenting procedures performed each year. The proper study of the management of carotid stenosis would start by identifying all persons with the condition, looking forward in time to identify the person who underwent a procedure and those who did not, and then determining whether the right decision was made. However, such a study has not been performed because there is no simple method for identifying the population with this condition.

Selected Publications, Posters, and Presentations

  • Mowery DL, Franc D, Ashfaq S, Zamora T, Cheng E, Chapman WW, Chapman BE. Developing a knowledge base for detecting carotid stenosis with pyConText. AMIA Symp Proc. Washington DC. 2014.
  • Mowery DL, South BR, Garvin J, Franc D, Ashfaq S, Zamora T, Cheng E, Chapman BE, Keyhani S, Chapman WW. Adapting a Natural Language Processing Algorithm to Support Stroke Cohort Generation. HSR&D/QUERI National Day. July 2015 (accepted)
  • Mowery DL, South BR, Madden E, Chapman BE, Chapman WW, Keyhani S. Filtering Negative for a Comparative Effectiveness Study of Stroke. AMIA 2015 Summits on Translational Bioinformatics. San Francisco, CA. Late Breaking Research Presentation.
  • PI: 
    Wendy Chapman