Mining social media messages—tweets, blogs, and Facebook posts—for health and drug-related information is of significant interest in what is called “pharmacovigilance” research.
Social media sites like Twitter have been used to monitor drug abuse and adverse reactions to drug use, as well as analyzing how people feel about drugs. However, most of these studies are based on aggregated results from a large population rather than specific sets of individuals.
Conducting studies at an individual level or specific groups of people requires identifying posts mentioning intake of medicine by the user.
Like what you’re reading? Stay ahead of your field and sign up here for our Build Your Career or Computing Edge newsletters to get content like this delivered right to your inbox weekly.
To this end, researchers from industry and academia developed a classifier to identify mentions of personal intake of medicine in tweets. They trained a stacked ensemble of shallow convolutional neural network (CNN) models on an annotated dataset.
The system produces state-of-the-art results, with a micro-averaged F-score of 0.693, according to the researchers.
The classifier has promising applications in the areas of psychology, health informatics, pharmacovigilance, and affective computing for tracking moods, emotions, and sentiments of patients who take medication and talk about it on social media.
Read more in the article “Detecting Personal Intake of Medicine from Twitter.”