"You may have lung cancer. Please consult a physician." That'd very likely be the worst alert you could ever see on your phone, right? In the future, though, it could be a reality. By looking at anonymous search history and cross-referencing it with demographic data, scientists from Microsoft Research propose that 39 percent of oncology diagnosis could be made a year earlier -- no Watson required. It definitely sounds like something out of Minority Report, but applied to healthcare.
Your search history could be an early detector for lung cancer
Pre-cognition, but for oncology instead of high crime.
Sponsored Links
Using Bing (this is Microsoft we're talking about, after all), the researchers looked for queries like "I was just diagnosed with lung cancer," and then worked backwards from there, according to Bloomberg. Then the scientists scoured those users' histories for symptom-related searches like bronchitis or chest pain. From there, things like smoking or exposure to radon gas were derived from location data and other search terms.
Of course, with this sort of thing, there's always the chance for false diagnosis. That 39 percent success rate is with one false positive per 1,000 patients. And dropping to one false diagnosis in 100,000 patients could still give three percent of folks an early warning.
Since cancer diagnosis often comes when a patient is already terminal, this feasibility study, with additional testing and research, could prove to be incredibly important. And it'd probably result in one notification you wouldn't absentmindedly swipe away.
