User Identification from Time Series of Fitness Data

Abstract

We explore the threat posed by disclosure of personal fitness information collected by wearable devices. In particular, we study a scenario where an attacker has a list of aggregated records produced by a group of users, which are stored as time series of steps and calories. We introduce a machine learning-based approach to identify one target person in the aggregated data while being in possession of other records from that person. We estimate how accurately an attacker can find the target's data when aggregated with other users by testing our approach on two public datasets. Our results show that personal fitness data possess identifying capabilities that should be accounted when they are shared or disclosed. Had a great day today! Here are my results: steps: 18134, calories: 2159, 2 workouts, and 8 hours sleep.20 3 45 Bob @bob89 19h aggregated database public data on social networks Eve # Day Steps Calories