Cellphone data mined to create personal profiles
Analyzing calls, text, and other cellphone data yields clues about a user’s personality, behavior, and even health status
As we live our lives, we leave behind a vapor trail of information: the places we go and how fast we get there, the people we talk to, the stores where we shop, taverns where we drink, churches where we pray.
Powerful computers in our purses and pockets are now recording that data. The cellphone is not just a communication device; it is a diary. And with the data from millions of these diaries, businesses, government agencies, and scientists are learning how to forestall medical crises, identify emotional problems, prevent the spread of infections diseases, or simply sell more pizza.
For Anmol Madan, phone data offer an extraordinary opportunity to keep people healthier at lower cost.
“There’s a fundamental problem in health care in that there’s almost no information about how patients behave when they’re outside the clinic,” said Madan, who holds a doctorate from the Media Lab at the Massachusetts Institute of Technology.
Madan founded Ginger.io, a company whose software turns a smartphone into a medical monitor. The app uses the phone’s GPS location chip to constantly track the user’s location. At the same time, it employs the phone’s accelerometer chip, which rotates the on-screen display when you turn the phone on its side, as a motion detector that keeps track of the user’s movements. It is a simple way to see if he or she is getting enough exercise.
In addition, the Ginger.io app logs the number and duration of phone calls. There is no threat to privacy; the software does not track who the user calls or what is said.
So why bother? Because of research showing that people who are more sociable also tend to be healthier.
“You want to make sure that somebody who’s got diabetes, that he’s not depressed,” said Madan.
So Ginger.io tracks a person’s phone usage over time. If a patient’s calling pattern radically changes, it could indicate an unwelcome mood shift, and the software will notify a physician. Madan said the technology could reduce hospital readmissions of chronically ill patients.
Meanwhile, scientists at the Massachusetts Institute of Technology and Harvard University say they can figure out the kind of person you are, simply by studying the way you use your cellphone. Between March 2010 and June 2011, researchers collected data on 69 study participants, using an app that recorded various indicators of how each phone was used.
“Do you call a lot of numbers, or all the time the same numbers? Do you call the same number every day at noon? Do you respond very quickly to a text?” said Jordi Quoidbach, a postdoctoral researcher in psychology at Harvard who was recently appointed an assistant professor at Pompeu Sabra University in Barcelona.
He and other members of the team also measured variables such as the locations from which calls and texts were sent, and how often the person used the phone from home.
“The way people interact with each other shows their personalities, even when they’re on a cellphone,” said Alex Pentland, director of MIT’s Human Dynamics Laboratory.
Each participant was also given standard psychological tests that identify personality traits. The researchers found the cellphone data enabled them to predict how people would perform on the tests. For example, they could spot an extrovert by tracking how often the phone is used at home, the distance traveled each day, and how many places are visited.
Yves-Alexandre de Montjoye, an applied mathematician at the MIT Media Lab who worked on the study, hopes it will be possible someday to run psychological tests on entire cities or countries, simply by analyzing the citizens’ phone records.
“Our main goal is both to show that it is possible . . . and to see if this method can be applied on a larger scale,” Montjoye said.
For example, scientists might use phone records someday to estimate how many Americans are prone to depression, with no need to conduct millions of individual tests.
But Quoidbach admitted the technique could be misused by marketing companies trying to gain insights into consumer buying habits.
“Right now it’s kind of a gray zone,” said Quoidbach, who favors strong safeguards to ensure that cellphone profiling does not infringe on privacy rights.
Such data, though, are already in commercial use. MIT’s Pentland cofounded a company, Sense Networks, to analyze usage data from smartphones to precisely tailor mobile advertising campaigns. The company’s clients include QIP Holder LLC, owner of the Quiznos chain of fast-food restaurants.
Sense Networks uses the GPS bread crumbs recorded by the cellphone to track where users go in daily routines — to the supermarket, the mall, nightclubs, museums, or movie theaters. That can indicate whether the cellphone user is a sensitive patron of the arts, a shot-and-a-beer sports fan, or both.
Sense Networks aggregates the data to create customized market segments — music lovers or sports junkies, for example. When someone in a targeted group launches a smartphone app, Sense Networks makes sure they will see a relevant advertisement.
Verizon Wireless does much the same thing with its Precision Market Insights program, which combines phone-location data with information about the apps running on customers’ phones and the websites they visit. The results are like a road map for marketers. During last year’s Super Bowl, for example, Precision Market Insights found more smartphone users visited the websites of Chevrolet, Best Buy, and Ford after their ads ran during the game — instant evidence the commercials had an impact.
Nathan Eagle, another MIT Media Lab veteran, wants to use cellphone data from millions of users to prevent the spread of infectious diseases.
Eagle is the chief executive of Jana Mobile Inc., a Boston company that conducts surveys and marketing campaigns via cellphones. Eagle and his wife, Caroline Buckee, an epidemiologist at Harvard School of Public Health, have used cellphone records of 15 million people in Kenya to track the spread of malaria in that country.
By studying cellphone records, scientists “get a real-time lens into how the aggregate population is actually behaving,” Eagle said. “That’s useful for anything from urban planning to public health.”
For example, IBM Corp. used cellphone location data to track commuters in Abidjan, the largest city in the African nation of Cote d’Ivoire, and plotted more efficient routes for bus lines to reduce commute times. IBM hopes to apply the same method in other cities.
Tracking people through their phones is bound to alarm privacy advocates, especially since the recent revelations that the National Security Agency has been secretly collecting vast amounts of telephone and e-mail data.
But public outrage probably will not halt the mining of mobile phone information.
“I’d love to be able to access the data that my [cellular] operators have on me,” Eagle said. “I think that data can be used in all sorts of great ways that goes far beyond marketing.”
After all, it is one of the most valuable databases on earth, and it gets more valuable whenever one of us switches on a cellphone.