ARGUS: Interactive Visual Analytics Framework for the Discovery of Disruptions in Bio-Behavioral Rhythms

Abstract

Human Bio-Behavioral Rhythms (HBRs) such as sleep-wake cycles and their regularity have important health ramifications. Smartphones can sense HBRs by gathering and analyzing data from built-in sensors, which provide behavioral clues. The multichannel nature (multiple sensor streams) of such data makes it challenging to pin-point the causes of disruptions in HBRs. Prior work has utilized machine learning for HBR classification but has not facilitated deeper understanding or reasoning about the potential disruption causes. In this paper, we propose ARGUS, an interactive visual analytics framework to discover and understand HBR disruptions and causes. The foundation of ARGUS is a Rhythm Deviation Score (RDS) that extracts a user's underlying 24-hour rhythm from their smartphone sensor data and quantifies its irregularity. ARGUS then visualizes the RDS using a glyph to easily recognize disruptions in HBRs, along with multiple linked panes that overlay sensor information and user-provided or smartphone-inferred ground truth as supporting context. This framework visually captures a comprehensive picture of HBRs and their disruptions. ARGUS was designed by an expert lead goal-and-task analysis. To demonstrate its generalizability, two different smartphone-sensed datasets were visualized using ARGUS in conjunction with expert feedback.