Internet searches lead to discoveries of drug side effects

Software that analyzed Internet searches was able to detect side effects caused by prescription drugs.
Written by Laura Shin, Contributor

Scientists at Microsoft, Stanford and Columbia Universities have found side effects to prescription drugs before the Food and Drug Administration knew of them ... all by analyzing Internet searches.

According to the study, published in the Journal of the American Medical Informatics Association on Wednesday, the researchers used automated software to sift through the searches of six million Internet users in 2010. They looked for queries relating to an antidepressant called paroxetine and a drug to lower cholesterol, pravastatin, and were able to conclude that combining the two caused high blood pressure.

The software uses data-mining techniques similar to those that power Google Flu Trends, which indicates the prevalence of the flu in locations based on search data.

The FDA's reporting system for drug side effects, the Adverse Event Reporting System, depends on doctors to notice a side effect and report it.

The researchers recommended that the Internet search analysis be incorporated into the FDA's current system for tracking adverse drug effects. “There is a potential public health benefit in listening to such signals,” they wrote in the paper, “and integrating them with other sources of information.”

How the technology detected the side effects

The study was based on previous work by the laboratory of Stanford bioengineering professor Russ B. Altman in which they used software to explore the data in the FDA reports and discover side effects caused by drug-drug interactions.

In May 2011, they reported that in doing so, they detected an interaction of paroxetine and pravastatin, and concluded that the combination of the two increased the patient's risk of developing hyperglycemia.

Dr. Altman then wondered if the same kind of conclusions could be reached in a more immediate manner -- using web searches. He enlisted Microsoft computer scientists to create a software toolbar that would collect and anonymize searches from users who allowed their search histories to be collected.

The result was 82 million searches that the lab group could analyze for searches such as paroxetine and pravastatin, as well as for terms related to hyperglycemia. It found that users who searched both drugs were also twice as likely to search for terms related to hyperglycemia, (i.e. “high blood sugar” or “blurry vision”) as users who only took one of the drugs.

“You can imagine how this kind of combination would be very, very hard to study given all the different drug pairs or combinations that are out there,” Eric Horvitz, a managing co-director of Microsoft Research’s laboratory in Redmond, Wash., told The New York Times.

Related on SmartPlanet:

via: The New York Times

photo: Root66/Wikimedia

This post was originally published on Smartplanet.com

Editorial standards