| Literature DB >> 18693990 |
David Carrell1, Diana Miglioretti, Rebecca Smith-Bindman.
Abstract
We coded 700 radiology reports from 373 women using an unmodified deployment of the Cancer Text Information Extraction System (caTIES), a publicly-available tool using natural language processing techniques. We were moderately successfully using caTIES for case ascertainment, successfully identifying 9/11 of a random sample of cancer case (sensitivity 82%) and 5/100 controls (specificity 95%) We are currently developing a classification scheme to assess clinical risk of ovarian cancer and identifying required extensions to caTIES algorithms.Entities:
Mesh:
Year: 2007 PMID: 18693990
Source DB: PubMed Journal: AMIA Annu Symp Proc ISSN: 1559-4076