Term variation and semantics for document classification and detection of obesity and its co-morbidities cases
Résumé
Within the translational medicine area, researchers aim at connecting more closely medical knowledge and individual patient data, in order to improve patient care quality. The 2008 I2B2 Obesity NLP challenge is to promote such research and to evaluate such systems within the same framework. We participated at this challenge. Our approach bets on use of semantics and term variants. The results obtained by our system are superior to the mean challenge results and seem to be promising, although there is room for improvements which we present and discuss in this paper.