Abstract
Patient discharge summaries provide detailed medical information about hospitalized individuals and a rich resource of data for clinical record text mining. The textual expressions of this information are highly variable. In order to acquire a precise understanding of the patient, it is important to uncover the relationship between all instances in the text. In natural language processing (NLP), this task falls under the category of coreference resolution. A key contribution of this paper is the application of contextual-dependent rules that describe relationships between co-reference pairs. To resolve phrases that refer to the same entity, we use these rules in two representative NLP systems: one based on the maximum entropy (ME) model, and the other being a system built on the Markov logic network (MLN) model. Our experimental results show that our proposed MLN-based system achieved an un-weighted average F-measure of 0.875. In this paper, we have described the main challenges in the resolution of co-reference relations within patient discharge summaries. Several rules were proposed to exploit contextual information and a total of two approaches were presented.