Paper: Improving Classification of Medical Assertions in Clinical Notes

ACL ID P11-2054
Title Improving Classification of Medical Assertions in Clinical Notes
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2011
Authors

We present an NLP system that clasifies the assertion type of medical problems in clinical notes used for the Fourth i2b2/VA Chalenge. Our clasifier uses a variety of linguistic fea- tures, including lexical, syntactic, lexico- syntactic, and contextual features. To overcome an extremely unbalanced distribution of asser- tion types in the data set, we focused our eforts on adding features specificaly to improve the performance of minority clases. As a result, our system reached 94.17% micro-averaged and 79.76% macro-averaged F 1 -measures, and showed substantial recal gains on the minority classes.