Mining Patients' Reviews in Online Health Communities for Adverse Drug Reaction Detection of Antiepileptic Drugs

In pharmacovigilance, the detection of adverse drug reactions is a task of utmost importance. This paper presents a data mining-based method to detect adverse drug reactions of anti-epileptic drugs from a dataset of patients' reviews collected from an online health community. The dataset is preprocessed and the unigram, bigram, and trigram are generated and then the adverse drug reactions of each anti-epileptic drug are extracted with the help of consumer health vocabulary and adverse drug reactions lexicon. Proportional reporting ratio is used to measure the association between each adverse drug reaction and antiepileptic drug. A list of ranked adverse drug reactions for each anti-epileptic drug is generated and validated against Drugs.com database. The results show the validity and utility of using patients' reviews in online health communities as a source for adverse drug reactions detection.