StandUp : an automated guide for improving physical activity of people using their handheld and desktop devices

Abstract

The number of desk jobs have increased phenomenally with the evolution of computers, especially in the last few decades. A modern day worker who is indulged into desk job spends nearly one third of her daily time sitting at work. Excessive sitting has been identi ed as a major health- risk factor and can lead to several health problems like diabetes, heart attack, cancer, increased mortality etc. Researchers have found out that sitting and physical activity are two separate behaviors and even regular exercising cannot negate the adverse a ects of excessive sitting. Taking regular breaks from sitting has been identi ed as the most e ective remedy to curb down the harmful e ects of sitting. This justi es the need of a ubiquitous system which tracks user activity and sitting time of a user and provide her with appropriate alerts and noti cations to guide her activity for a better and healthy lifestyle. Existing solutions include costly wearable devices and battery draining smartphone applications which are not prefered by the users. There also exists a category of desktop applications which alert users to take breaks when they are working on workstations, but the coverage of such systems is limited to the time spent by user on workstation only. We propose and develop a system StandUp which leverages on existing workplace infrastructure to provide sensing hints to detect user's presence which are complemented with the activity monitoring capabilities of a smartphone to track user's activity and sitting time throughout the day. It also issues alerts and noti cations to guide the user. StandUp system comprises three main components namely a mobile application, a desktop application and a cloud service which acts as a intermediary and also aggregates the activity related data coming from both desktop and mobile application. We developed a fully functional prototype system which includes desktop application build on top of .NET framework for workstations running Windows OS, native mobile application for Android smartphones and a django based cloud service which runs on top of an Apache server. We also give an elementary evaluation of the system and present some interesting statistics and activity patterns from the data collected by our system for nearly four weeks among few users. We also record the compliance of the noti cations issued from the system by the users. We also evaluate our triggered sensing approach and present the daily tracking coverage of the system for a single user over a span of nearly two weeks. We compare the power characteristics of our mobile application with a well known activity tracking application available on Google Play Store. We plan to further improve the system and aim to build a more robust and e ffective system, and deploy it on a larger scale , which we scope as our future work.