The FDA has approached Google for help identifying adverse drug events in search data. The regulator hooked up with Google to access the insights of Evgeniy Gabrilovich who has published papers on the subject of mining search queries for adverse event reports.
The meeting, which came to light when the FDA posted a record of the event on its website was described as an introductory session to give the FDA an opportunity “to begin a discussion on how we might collaborate with Google on identifying adverse event data, using Google’s technologies and data.” Neither the FDA nor Google is providing further details at this time although the identity of Google’s representative on the call and the FDA’s increasing interest in mining new sources of data in its pharmacovigilance activities give clues about the likely focus of the meeting.
In 2013, Gabrilovich co-authored a paper about using 176 million Yahoo search queries to spot adverse events. The paper was published in the Journal of Medical Internet Research.
Gabrilovich and his collaborator looked for searches that paired common symptoms–such as “cramps” and “tired”–with the name of a drug. The mining led to previously unknown links between drugs and adverse events. Specifically, the researchers claimed their approach was good at finding adverse events that emerged long after the start of treatment, a subsection of drug-related side effects that traditional methods may be poor at identifying. Whether this is a truly useful addition to the pharmacovigilance armory remains to be seen, but the FDA is keen to explore its options.
The meeting with Google took place the week before PatientsLikeMe revealed it had signed a research collaboration with the FDA. And Microsoft has been working informally with the regulator for years. Collectively, the relationships with tech companies attest to the FDA’s interest in making use of new sources of data.