A new system will be able to tell whether you have really worked out or not

Cheat-proof activity trackers expose couch potatoes

Patients trying to abuse activity trackers to fake activity will hit a wall with a new algorithm designed to spot cheaters. 

Developed by researchers from Northwestern University in Chicago, the system will help insurance companies to separate the wheat from the chaff and waste less money rewarding patients that don’t really work out. Similarly, it could be used by doctors to monitor whether patients are following their advice or not.

"As healthcare providers and insurance companies rely more on activity trackers, there is an imminent need to make these systems smarter against deceptive behaviour," said Sohrab Saeb, lead author of the study published in the latest issue of the journal PLOS ONE. "We've shown how to train systems to make sure data is authentic."

The researchers opted for a different approach from what has been used previously; instead of training their system to recognise genuine activity, they trained it to recognise cheating. This slight change of angle rendered surprising results. While a system trained on real activity was able to filter out only 38 per cent of cheaters, the new technique was accurate in 84 per cent of cases.

The system can detect, for example, when the cheater simply shakes the phone or swings it in his hand, to pretend he or she is walking, while actually lying on a couch.

The system, which could work in smartphones as well as fitness wristbands and watches, learns to recognise when one person cheats and applies this knowledge to everyone else displaying the same behaviour.

"Very few studies have tried to make activity tracking recognition robust against cheating," said Konrad Kording, research scientist and associate professor in physical medicine and rehabilitation. "This technology could have broad implications for companies that make activity trackers and insurance companies alike as they seek to more reliably record movement."

However, the researchers admit, there is a way to trick the device – to give it to someone else, who is actually performing the activity, or simply attach it to a dog’s collar.

 

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them

Close