[home] [Personal Program] [Help]
tag Human Behaviour Monitoring Through Smartphones
Kostas Konsolakis, Hermie Hermens, Oresti Banos
Session: Poster session II
Session starts: Thursday 24 January, 16:00



Kostas Konsolakis (University of Twente)
Hermie Hermens (University of Twente)
Oresti Banos (University of Granada)


Abstract:
The recent technological advances have enabled new potentials for monitoring people’s health. Human behaviour analysis through smartphones has been an active field for more than a decade, playing a major role in promoting active and healthy ageing. Smartphone sensors, such as accelerometer and gyroscope, but also GPS, Wi-Fi, and Bluetooth, can detect users’ activity and movement (for example sitting, walking, running) [1]. Data related to screen touch events and users’ response time can be used to track cognitive states, such as attention and alertness [2]. Phone calls and text messages have been used to monitor users’ social life [3], while audio and microphone signals, combined with the users’ physical activity, can be used to detect mood (boredom, happiness, anxiety, etc.) [4]. Even though human behaviour can be detected through physical, cognitive, emotional and social sensing, there is not a clear investigation for detecting human behaviour in a holistic approach, using smartphone data. The present study aims to investigate human behaviour through smartphone data, by answering the following research question “Could a smartphone monitor human behaviour in a broad sense?”. For the data acquisition phase, we will conduct an experiment in both a controlled and uncontrolled environment. Subjects will be asked to participate voluntarily and perform a series of consecutive tasks. Four main scenarios will be investigated by examining different classification approaches, related to users’ physical activities, social interaction, cognitive abilities and emotional states. Based on the collected data from the controlled experiment the model will be trained to predict momentary behaviours in a short-term period, while data from the uncontrolled environment will be used to detect behaviour routines and changes over a longer period of time.