Big data: Bing can (sometimes) tell if you have pancreatic cancer

Microsoft researchers have written a paper about how search engines can warn people of disease and physical illness, based on their search history.

The team managed to build a model that can predict if users have pancreatic cancer. Unfortunately, it has an identification rate of just 5 to 15 percent but the false positive rate is as low as one in 100,000.

The model was created by using anonymized data from Bing. First, the researchers identified users with pancreatic cancer by mining the search data for queries that indicated someone had recently been diagnosed. Then the team worked backward, looking for earlier queries that could serve as warning flags.

These kind of big data tools can be helpful instruments to improve people's lives as early screening can increase survival rates.

While five-year survival rates for pancreatic cancer are extremely low, early detection of the disease can prolong life in a very small percentage of cases. The study suggests that early screening can increase the five-year survival rate of pancreatic patients to 5 to 7 percent, from just 3 percent.

The big question now is how Microsoft is going to proceed with this. The software giant declined to offer specific details but given the recent creation of a Health & Wellness division at Microsoft it seems we can expect more health-related tech or perhaps an online health service in the future. Full details at NY Times.

By the way, this isn't the first time Microsoft used search data to improve healthcare. In 2013, the company managed to detect unreported side effects of prescription drugs before they were found by the FDA's warning system.