VISUAL ANALYSIS OF MOBILE SENSING TIME-SERIES DATA: IDENTIFYING INDIVIDUAL AND RELATIVE BEHAVIOURAL PATTERNS

Abstract

Mental well-being is increasingly demanded due to growing concerns about mental health. At the same time, the Internet and smartphones are transforming the world in unprecedented ways. This pervasiveness opens up new avenues for research by providing access to an individual’s behaviour and daily habits. Unobtrusive data collection and analysis from smartphone sensors is a promising approach to address- ing mental health issues and have been the focus of many research studies. In this work, we explore this opportunity by analyzing data collected from smartphone usage and leveraging the advantages of data visualization and machine learning methods to possibly identify and compare behavioural indicators and patterns that can indi- cate mental health. We developed a visualization system to interact with extracted features about behavioural indicators like screen usage, calling, and sleep to assess the daily routine of participants under study. We also present two usage scenarios to demonstrate our visual approach’s applicability in exploring the given dataset.