Semantic Information in Medical Information Systems: Utilization of Text Mining Techniques to Analyze Medical Diagnoses
Andreas Holzinger, Felix Mödritscher, Regina Geierhofer, Roland Tatzl
Abstract
Most information in Hospitals is still only available in text format and the amount of this data is immensely increasing. Consequently, text mining is an essential area of medical informatics. With the aid of statistic and linguistic procedures, text mining software attempts to dig out (mine) information from plain text. The aim is to transform data into information. However, for the efficient support of end users, facets of computer science alone are insufficient; the next step consists of making the information both usable and useful. Consequently, aspects of cognitive psychology must be taken into account in order to enable the transformation of information into knowledge of the end users. In this paper we describe the design and development of an application for analyzing expert comments on magnetic resonance images (MRI) diagnoses by applying a text mining method in order to scan them for regional correlations. Consequently, we propose a calculation of significant co-occurrences of diseases and defined regions of the human body, in order to identify possible risks for health.