MIT's Technology Review has showcased a joint research effort by Technion doctoral student Kira Radinsky and Microsoft researcher Eric Horvitz, who developed a program that at its current stage has successfully predicted future disease and violence outbreaks with 70 to 90 percent accuracy. The prototype, which was developed in the U.S., merged two decades worth of New York Times archives and other Internet sources such as Wikipedia to make its predictions.
“I truly view this as a foreshadowing of what’s to come ... eventually this kind of work will start to have an influence on how things go for people,” Horvitz told Technology Review.
Radinsky and Horvitz used archived news from 1986 to 2007 in their research to analyze multiple media outlets and make predictions based on those reports. The software successfully predicted that there would be an outbreak of cholera in Angola in 2006, based on compiled data that showed that years of drought, which Angola was suffering at the time, were usually followed by cholera outbreaks. According to the report, the program successfully predicted a cholera outbreak in Angola again in 2007.
"A cholera outbreak warning can be given a year in advance," Radinsky told MIT. She explained the program worked "retrospectively — looking at past events to predict future outbreaks."
According to Horvitz, while the world may have changed in recent decades, human nature and many aspects of the environment have not, and compiling older information was still very pertinent to making accurate predictions.