PHEE: A Dataset for Pharmacovigilance Event Extraction from Text

<p>The PHEE dataset contains over 5,000 finely annotated pharmacovigilance events from public medical case reports. Two types of events, the adverse events and the potential therapeutic events, are annotated.&nbsp;For each event, we annotate the event trigger and hierarchical arguments.&nbsp; The main arguments (coarse-grained spans) include subject, treatment and effect. Further fine-grained sub-arguments -&nbsp;<em>age</em>, <em>gender</em>, <em>race</em>, <em>number of patients</em> (labelled as population) and <em>preexisting conditions</em> (labelled as subject.disorder) for the subject argument and&nbsp;<em>drug </em>(and their combinations), <em>dosage</em>, <em>frequency</em>, <em>route</em>, <em>time-elapsed</em>, <em>duration</em>,<em> target disorder</em> (labelled as treatment.disorder) for the treatment argument -&nbsp;are then annotated upon main arguments.</p> <p>We provide two formats of data: visualisation-friendly brat-format data and structured json data for the convenience of use.</p> <p>&nbsp;</p>