Stress data visualization

Abstract

Stress related research is becoming nowadays increasingly popular. Special attention is devoted to stress at work. Recent studies provide proofs of dangerous influence that stress has on human health. It leads to the appearance of a new important part of self-healthcare: stress monitoring. Existing ways of stress monitoring include various ways of data collection, visualization and representation to the user. The focus of the current work is on the second part: stress data visualization techniques. One of the problems of existing visualization techniques for stress level is isolation of stress data from the real life of a person, experiencing emotional arousal. It is not always easy to associate some specific feature showed in the visualization with an event taking place at that particular moment of time, since it requires good memory from the user. However, in context of stress experienced at work, this should be possible, because work time usually has some structure and events are often scheduled for specific time moments. In this work, we try to put together two sources of important information: stress level data of a user and a personal calendar, containing information about professional and personal life events. After taking a look at existing techniques for visualization of time-series data, we discuss how some of these techniques can be adapted for representation of stress level of a person. We propose a number of new ideas for stress data visualization, and possible ways of combining it with calendar information. We discuss different alternatives of stress data aggregation by time (per minute, half a day, and one day) and by event types, embedding of the visualizations into the calendar views (arrangement by day, week, and month), variations on number of users shown on a visualization. Several alternatives are validated in a small user study. The validation results prove that there is a visualization technique prefered by most users. To gain a better understanding of user view on the month visualization, another study with a small group of participants is performed. The results of the study show differences in user perception depending on the visualization technique, even if the same piece of data is visualized; some similarities in men and women perception of visualizations; existence of different groups of users having the similar view on the questions and factors defining the choice of each group. To prepare the use of validated visualization ideas in real applications, we investigate possible ways of storage and management of sensor stress level data. Considered alternatives are evaluated in terms of performance, ease of access and processing data, maintainability, and memory use, and conclusions about the most appropriate way of data organization are made.

Date of Award	28 Feb 2014
Original language	English
Supervisor	Natalia Sidorova (Supervisor 1), R.D. Kocielnik (Supervisor 2) & M.A. Westenberg (Supervisor 2)