Detection and analysis of medical misbehavior in online forums
Résumé
Social media is an important source of information on behaviour and habits of users. It has been used as such in public health research to monitor adverse drug effects and drug misuse among others. We propose to study drug non-compliance in health online forums. First, we use supervised classification to detect non-compliance messages and obtain 0.436 of F-measure. Then, we manually analyse the content of the messages to learn what kinds of behaviour can be detected, and to study the effect the social media can have on patient's compliance behaviour.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...