We help our customers to acquire and analyse data from wearable movement sensors, for example, to automatically detect the amount of daily physical activity and the types of physical activities.
With the VTT algorithms,
we can analyse raw, real-world movement sensor data, for instance from an
accelerometer, in order to detect several types of physical activities such as lying
down, sitting, standing, walking, running, bicycling, rowing, playing football,
sleeping, etc. We are offering the existing algorithms, software applications
for data annotation and visualization, and know-how in wearable sensing, signal
analysis, algorithm development, validation studies and software development.
Unlike many traditional
classifiers, the VTT algorithms have a low computational cost, and are thus also
suitable for embedded use, for example in mobile phones and other wearable
devices. Our algorithms and software have been developed in several scientific
studies. Our publications have been widely cited and are among the
top-publications in the world.
In addition to activity
types, more detailed analysis of, for example the steps or arm movements, is
possible, as well as an estimation of total energy expenditure, based on data
from unobtrusive, wearable sensors. We have, in addition to data analysis, worked
on the selection of sensors for validation studies, the design of a data
collection protocol as well as with the data collection. We are also developing
software applications for data annotation and visualization.
The algorithms have been
developed using the carefully collected and well-annotated data libraries from
several research studies. We have developed algorithms for
customers in several projects. Our customers
include Firstbeat Technologies and Nokia.
Where can we apply motion
The algorithms can be
used in different application domains, for instance in self-monitoring, preventive care,
rehabilitation programs, occupational health, stress studies, exposome studies,
etc. The inferred information can be used to profile users, to coach the users,
and so on.
In our projects, we can utilize
the existing know-how, algorithms, software applications for data annotation
and visualization, and the facilities of a Personal Health Systems laboratory
with, for example, treadmill, exercise bike, rowing machine and devices for
The work on monitoring
physical activities was started in a national research project named Palantir
“Context sensing with wearable sensors and data fusion”. Currently this work is
applied in the European project called PredictND for studying possibilities of
using movement data in prediction of dementia www.predictnd.eu.
“Activity classification using realistic data from wearable
sensors”; Pärkkä, Juha; Ermes, Miikka; Korpipää, Panu; Mäntyjärvi, Jani; Peltola,
Johannes; Korhonen, Ilkka. IEEE Transactions on Information Technology in
Biomedicine. Vol. 10 (2006) No: 1, 119–128
“Detection of daily activities and sports with wearable sensors in
controlled and uncontrolled conditions”; Ermes, Miikka; Pärkkä, Juha;
Mäntyjärvi, Jani; Korhonen, Ilkka. IEEE Transactions on Information Technology
in Biomedicine. Vol. 12 (2008) No: 1, 20–26
“Estimating intensity of physical activity: a comparison of
wearable accelerometer and gyro sensors and 3 sensor locations”; Pärkkä, Juha;
Ermes, Miikka; Antila, Kari; Gils, Mark van; Mänttäri, Ari; Nieminen, Heikki. Proceedings of the 29th Annual
International Conference of the IEEE Engineering in Medicine and Biology
Society. Lyon, France, 23–26 August 2007. IEEE Engineering in Medicine and
Biology Society (2007) , 386