查看文章

In recent years the interest for performing research on biomedical and clinical data within language technology has increased immensely. There are many reasons for this. For instance, such domain specific data contains vocabularies and language use that is very interesting and not previously studied from a linguistic point of view. Also, such data contains a potentially large amount of information that could be useful for other research areas such as Medical Informatics, Epidemiology and Biomedicine, to mention only a few. However, research on clinical data is still very limited, since many privacy issues need to be solved and many ethical aspects need to be taken into account when working with Electronic Patient Records (EPRs). In order to obtain access to clinical data, issues concerning integrity and privacy need to be properly secured. Moreover, the data needs to be fully deidentified.

This paper describes some general characteristics of a large corpus of EPRs written in Swedish, which our research group plans to use for further research. We believe such research will be very interesting within the area of for instance Information Access, Text Mining and Medical Informatics. The corpus has been granted access by the hospital management from which the corpus is derived after approval from the Regional Vetting Board.