Dynamic generation of a health topics overview from consumer health information documents

Online health information use is increasing, but can be too difficult for consumers. We created a system that dynamically generates a health topics overview for consumer health web pages that organises the information into four consumer-preferred categories while displaying topic prevalence through visualisation. It accesses both a consumer health vocabulary and the Unified Medical Language System (UMLS). We evaluated its ability by calculating precision, recall, and F-score for phrase extraction and categorisation. We tested pages from three different consumer web sites. Overall, precision is 82%, recall is 75%, and F-score is 78%, and precision between sites did not significantly differ.