Article ID Journal Published Year Pages File Type
4961703 Procedia Computer Science 2016 7 Pages PDF
Abstract

In health care information systems, electronic health records are an important part of the knowledge concerning individual health histories. Extracting valuable knowledge from these records represents a challenging task because they are composed of data of different kind: images, test results, narrative texts that include both highly codified and a variety of notes which are diverse in language and detail, as well as ad hoc terminology, including acronyms and jargon, far from being highly codified. This paper proposes a combined approach for the recognition of named entities in such narrative texts. This approach is a composition of three different methods. The possible combinations are evaluated and the resulting composition shows an improvement of the recall and a limited impact on precision for the named entity recognition process.

Related Topics
Physical Sciences and Engineering Computer Science Computer Science (General)
Authors
, , , , , , ,