The ADE Discoverer is an initiative to harvest adverse drug effects (ADEs) casually reported in forum discussions, user reviews, and other social media. Initially, the Discoverer will focus on a limited number of health-care related websites;however, as Phase I progresses there will be a focus on implementing techniques of automatic pattern recognition to allow structured data extraction from additional unknown websites. Once relevant social media is extracted, the Discoverer must then identify any occurrences of adverse drug effects. This is particularly challenging in forum discussions, where members are often discussing well known side effects, or their distinct experiences with multiple drugs in the context of a single post. In order to recognize which adverse effect terms were actually experienced as the result of the commenter taking a drug, a series of machine-learning algorithm based classifiers have been proposed. The classifiers will create classification models based on training data provided by the manual identification of ADEs over thousands of human reviewed posts. In Phase I and beyond, a constant focus will be on improving the ability to recognize human-experienced ADEs in social media. The side effects that pass through the classification filter will be correlated with FDA-reported side effects in order to power the ADE Search and Analysis Workbench. Using the interactive ADE Workbench, researchers will be able to quickly gain the insights necessary to make informed decisions about medications. Publicly available components of the workbench include: total side effect reports by drug, drill-down functionality to view raw, relevant social media and medical reports, and a drug comparison application to compare similar drugs effectiveness and likelihood of side effects side-by-side.
In the 2010 Omnibus Solicitation the Center for Drug Evaluation and Research has identified several research and development opportunities that directly correspond to the work accomplished by the ADE Discoverer. Among the most relevant opportunities include: """"""""Develop methods for timely active surveillance of newly approved drug products in large populations to identify both expected and unexpected outcomes."""""""", and """"""""Develop methods for actively collecting information on all cases of classically drug-associated events to augment the FDA's current passive surveillance system."""""""" The ADE Discoverer actively monitors massive amounts of user generated content to identify adverse drug effects that would otherwise be missed by the FDA's current passive surveillance system.