Opportunistic Discovery of Personal Places Using Smartphone and Fitness Tracker Data

摘要

It is becoming increasingly important to accurately detect a user's presence at certain locales (such as workplace, home, fitness studio, parks, etc.) during certain times of the day, e.g., to study a user's mobility, health, and fitness behavior or social interaction patterns and to enable the delivery of targeted services. These personal places of significance can be determined using segmentation of location traces into a discrete sequence of places. However, location traces often suffer from data gaps, especially indoors, and this may lead to a large number of small and incomplete segments, where many of these segments actually belong together. Recent developments in health and fitness tracking make it possible to continuously collect and analyze a user's biometric data (such as step counts, calorie burn, and heart rate), including during times when location data may be missing. This opens the opportunity to utilizing biometric data to verify a user's presence at a specific place. Specifically, this paper proposes a novel segmentation approach that opportunistically fills gaps in a user's location traces (collected by the user's smartphone) using readily available, coarse-grained, minute-level biometric data collected from a user's health or fitness wearable. In our analysis of more than 450 subjects' data from a two-year long mobile health study, we demonstrate that our approach yields fewer, but more complete and accurate segments than state-of-the-art approaches.