Automatic Extraction of Personal Experiences from Patients' Blogs: A Case Study in Chronic Obstructive Pulmonary Disease
Mark Greenwood, Glyn Jones Elwyn, Nick A Francis, Alun D. Preece, Irena Spasić · 2013
People with long-term illness such as chronic obstructive pulmonary disease (COPD) often use social media to document and share information, opinions and their experiences with others. Analysing the self-reported experiences of patients shared online has the potential to help medical researchers gain insight into some of the key issues affecting patients. However, the scale of health conversation taking place online poses considerable challenges to traditional content analysis. In this paper, we present a system which automates extraction of patient statements which refer to a personal experience. We applied a crowd sourcing methodology to create a set of 1770 annotated sentences from blog posts written by COPD patients. Our machine learning approach trained on lexical features successfully extracted sentences about patient experience with 93% precision and 80% recall (F-measure: 86%). Automatic annotation of sentences about patient experience can facilitate subsequent content analysis by highlighting the most relevant sentences to this particular problem.