• This field is for validation purposes and should be left unchanged.
Guidance and results for the data-rich, yet information-poor.

Data Mining Uncovers Drug Interactions

medicine Uncovering hidden connections among data is what data mining is all about. Extracting information from vast stockpiles of data can be a daunting task, but is always one worth undertaking.

The challenge was certainly worthwhile in the case of a group of scientists from Microsoft, Stanford, and Columbia University. For the first time, unreported side effects of various medications were able to be detected and classified before the Federal Drug Administration’s warning system could find them.

By using automated software, the scientists were able to troll millions of internet queries for similarities. The study focused in particularly on an antidepressant and a cholesterol-lowering drug, which were ultimately found to cause a dangerous spike in patient blood sugar upon interaction. Prior to this study, the only way such an interaction would be noticed would be if a physician detected the result in a patient and submitted their finding through the FDA’s Adverse Event Reporting System.

This project came to be when Russ B. Altman, chairman of the bioengineering department at Stanford, began work to automate searches of drug-drug interactions within the FDA’s reports. Upon identifying the interaction within that source, Altman wondered if there might not be a more accurate, faster way to make such connections. And so, he turned to Microsoft.

Microsoft engineered a software toolbar to scan anonymized data collected from web users who allowed their browser information to be shared. Through this method, the group was able to analyze 82 million individual searches for drug, symptom, and condition queries. After continually refining the searches, the group was able to glean that those individuals who searched for both of the drugs in question within a twelve month period were much more likely to search for symptoms indicating a hypoglycemic reaction.

As reported by the New York Times, the researchers were surprised by the strength of the “signal” that they detected in the searches and suggested that methods like this one would be a valuable tool for the F.D.A. to add to its current system for tracking adverse effects. “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,” said Eric Horvitz, a managing co-director of Microsoft Research’s laboratory in Redmond, Washington. “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.”

The FDA has begun to undertake similar methods of research, hoping to study the behaviors of up to 100 million American pharmaceuticals customers. Thanks to innovative and careful data mining, the FDA now has the tools to garner new information which will help to protect the American public from currently unknown risks.

If you’re interested in learning more about how Data Mining can help you unlock hidden information in your own field, consider joining The Modeling Agency for a free webinar. Or, for a more in-depth experience that will leave you with fuller mastery of this valuable tool, perhaps one of TMA’s training seminars is for you!

Data Mining Webinar

Learn How to Get Predictive Modeling
Off the Ground and Into Orbit
1 Hour Live Interactive Event

Why Train With TMA?

Determine whether TMA training is right for you, and learn why TMA is truly the best option for live classroom analytics training.