Google Search could diagnose unknown drug side effects

Google could soon be helping diagnose otherwise unknown drug side-effects. The search engine might be looking for a SEO manager to sort out its own Google rankings, bizarrely, but the Food and Drug Administration in the US is turning to the company's search powers to help it uncover otherwise elusive drug side-effects.

Bloomberg spotted the unusual pairing in the FDA's public meeting notes of 9 June, which show that senior Google research scientist Evgeniy Gabrilovich was in attendance for the "Google/FDA Adverse Event Trending Teleconference" -- "adverse event" referring to the unknown side-effects that only transpire once a drug is in widespread circulation. 

The idea is, with enough data, serious contenders for additional drug side-effects could be whittled down from among all the paranoid search entries the public type everyday. The FDA was beginning discussions with Google on "identifying adverse event data, using Google's technologies and data," Bloomberg reports FDA spokesman Chris Kelly as saying.

Kelly added to WIRED.co.uk: "The agency met with Google for a high-level, general discussion of their technology and data capabilities. The teleconference was for each side to introduce themselves and for the agency to begin a discussion on how we might collaborate with Google on identifying adverse event data using Google’s technologies and data. There will be a follow-up teleconference with Google in the future."

Google has a strong record of helping national institutions investigate matters of health, often in an emergent capacity or in preparation of one. In 2011 the search engine was proposed as a tool for tracking the spread of the drug-resistant MRSA bacteria, by analysing city-by-city search terms in real time. This came two years after Google looked at search data from Mexico for the Centres for Disease Control, and found it could have caught an outbreak of swine flu early. Twitter also has huge capacity for tracing disease spread and has been named in several studies looking into the efficacy of social media in predicting outbreaks.
Now, it seems the FDA wants Google to do its work for it. The agency is responsible for checking every drug in circulation in the US has been clinically tested on numbers large enough to get population-wide side-effect predictions. However, this is not an assurance that there are no side-effects outside of those already discovered. Additional side effects can be highlighted through official channels, but it's a slow and inefficient system. Creating a robust alternative that collates vast amounts of data on side effects could lead to them getting added to lists faster, or even alert the FDA to a side-effect that could potentially lead to a drug being stripped of safety approval.
Data miner Gabrilovich co-authored a paper in 2013 that looked into the viability of such a robust system. It argued that, rather than relying on "spontaneous reports by patients and health care providers" of side effects, which may only come about as a result of long-term use (not accounted for in clinical studies) or use alongside other drugs, a "low cost, fully automated method" that could be run continuously would be far superior. It would mean the FDA could look specifically for the different potential reasons behind the side-effects, including whether they are being taken alone or in conjunction with other common drugs.
While clinical trials generally look at a few hundred or thousand patients, Gabrilovich's experiment ran 176 million search results. He was able to validate the findings by also testing for common side-effects of particular drugs. In the end, he concluded that the most serious early-onset side effects would be more likely to be reported to and picked up by the FDA and others. But that "less acute later-onset ones are better captured in web search queries".

Two years later, it looks like the FDA is taking notice, and finally looking into collaborating with big tech to solve this unwieldy challenge.