Personalised health document summarisation exploiting Unified Medical Language System and topic-based clustering for mobile healthcare

Abstract

According to the growing interest in mobile healthcare, multi-document summarisation techniques are increasingly required to cope with health information overload and effectively deliver personalised online healthcare information. However, because of the peculiarities of medical terminology and the diversity of subtopics in health documents, multi-document summarisation must consider technical aspects that are different from those of the general domain. In this article, we propose a personalised health document summarisation system that provides a reliable personal health-related summary to general healthcare consumers via mobile devices. Our system generates a personalised summary from multiple online health documents by exploiting biomedical concepts, semantic types and semantic relations extracted from the Unified Medical Language System (UMLS) and analysing individual health records derived from mobile personal health record (PHR) applications. Furthermore, to increase the diversity and coverage of summarised results and to display them in a user-friendly manner on mobile devices, we create a summary that is categorised into subtopics by grouping semantically related sentences through topic-based clustering. The experimental evaluations demonstrate the effectiveness of our proposed system.