Clinical Narrative Parser Using Phrase Structured Grammar

Abstract

Clinical parsing is important for parsing of clinical narrative so that it will be converted into structured report which is understandable by everyone. In this paper we discuss about phrased structure grammar used by constituency parser to generate structured clinical sentences. Clinical reports differ for different types of diseases as well as for hospitals also because of using their specific report format. In this paper we focus on rules which are applicable for specific clinical reports. PCFG (probabilistic context free grammar) is used for calculating approximation of tree. Recall, Precision and bracketing f-measure is also calculated, when compared to Stanford parser generated tree. Based on medical domain clinical narratives can be extracted by using NLP systems.

Keywords: Clinical Narratives, Constituency Parser, Phrase Structure Grammar, Probabilistic Context Free Grammar, recall, precision, f-measure