Oliver Amft, Wearable Computing Lab., ETH Zurich
Gloriastrasse 35, CH-8092 Zurich, Switzerland

Detection of nutrition phases and the eating micro-structure using wearable
sensors. Talk given at pHealth2006: International workshop on wearable micro-
and nanosystems for personalised Health, Lucerne, Switzerland, January 31, 2006.


Abstract
We present in this talk a project aiming at continuously monitoring human
nutrition phases, detecting nutrition habits and providing information on
malnutrition to health professionals.

For this purpose different unobtrusive body-worn sensors are deployed. We
present relevant on-body sensing domains, appropriate acquisition principles and
deployed pattern recognition methods. We discuss the value of these sensing
domains for nutrition monitoring and briefly depict evaluation results. It is
envisioned that preprocessed information from such independent sensing domains
can be combined in a context recognition system that automatically monitors
complete nutrition events. The expected information derived from this system
include nutrition habits, e.g. rate of intake, duration and daily schedule, all
of which can be relevant health issues. We will show how such parameters can be
extracted from the sensing domains.

Moreover the recognition system will provide information on the category of the
nutrients. This information is valuable for a variety of applications, including
obesity management, drug intake monitoring and dehydration control. We expect
that this system in combination with personalised coaching from health
professionals will help significantly to ambulatory analyse different forms of
malnutrition that cannot be observed with established medical
methods. Consequently this system will - when continuously used - support the
user to achieve or maintain a healthy lifestyle.