Northwestern researchers have designed a way to teach smartphones to spot fake activity on smartphone and wearable activity trackers.

With health care providers and insurance companies looking to digital trackers to monitor patients and customers and reward healthy behavior, such as Cigna’s “Just Walk 10,000 Steps a Day” program, Northwestern Medicine and Northwestern Rehabilitation Institute of Chicago researchers set out to increase these trackers’ ability to detect forged activity.

“As health care providers and insurance companies rely more on activity trackers, there is an imminent need to make these systems smarter against deceptive behavior,” lead study author Sohrab Saeb, a postdoctoral fellow at the Feinberg School of Medicine’s Center for Behavioral Intervention Technologies, said in a news release. “We’ve shown how to train systems to make sure data is authentic.”

Researchers found that once a tracker learns to spot how one person cheats, it will recognize the same fabricating behavior in someone else. The tracking systems the researchers trained using falsified data were 84 percent accurate in spotting that behavior later, a substantial increase from the 38 percent accuracy of tracking systems trained using real fitness activity.

The study, funded by grants from the National Institute of Mental Health, a division of the National Institutes of Health, tested smartphone activity trackers, but could also be applied to wearable sensor-based activity trackers like Fitbit bracelets, researchers said.

The training method is not entirely foolproof, though, Saeb cautioned.

“If someone attaches the activity tracker to a dog, the system can’t recognize that,” he said in the release.

Email: [email protected]
Twitter: @maddycfox