![](bg6.png)
4 CONCLUSIONS
The final results using MLP neural networks for fall
detection have been quite satisfactory. The
application classifies correctly 92% of the validation
group falls, better performance than other detection
methods: 80% in (Chen et al., 2005). Moreover, the
number of false alarms is drastically reduced to 1%,
which leads to enhance users trust on the fall
detector. Nevertheless, a more extensive study with
more users being also elderly has to be developed in
order to gather more data and confirm the results.
Although the portable device can run for months
with the same battery, the system needs a computer
to analyze all the data. In order to reduce costs, it is
possible to analyze the pattern remotely. As the
amount of exchanged data is reduced, it could be
sent via ADSL (if the person is at home), GPRS or
even SMS to a service center. Anyhow our
application gets better performance than others
embedded in a microcontroller but a higher cost and
complexity. To overcome this, we are currently
minimizing the neural network size so it can run in a
microcontroller or FPGA.
ACKNOWLEDGEMENTS
This work was supported by the Spanish MCYT
under project Ambiennet (TIN2006-15617-C03-02)
and by the EU under projects MonAmi (IST-5-
0535147) and EasyLine+ .
REFERENCES
Alwan, M., et al., 2006. ‘A Smart and Passive Floor-
Vibration Based Fall Detector for Elderly’. Proc. The
2nd IEEE International Conference on Information &
Communication Technologies: from Theory to
Applications ICTTA’06, Damascus, Syria, April 2006.
Chen, J., et al., 2005. ‘Wearable Sensors for Reliable Fall
Detection’. Proc. 27th Annual International
Conference of the Engineering in Medicine and
Biology Society, IEEE-EMBS 2005, September 2005,
pp. 3551–3554.
Degen, T., et al, 2005. ‘SPEEDY: a fall detector in a wrist
watch’. Proc 7th IEEE International Symposium on
Wearable Computers, October 2005, pp. 184–187.
Del Hoyo Alonso, R. 2003. ‘Supervised classification with
Associative SOM.’ IWANN 2003: 334-341
Doughty, K., et al, 2000. ‘The design of a practical and
reliable fall detector for community and institutional
telecare’, Journal of Telemedicine and Telecare,
February 2000, 6, (1), pp. 150–154.
Eurostat web: http://epp.eurostat.ec.europa.eu/ last visited
2007
Geer, D., 2005. ‘Users make a Beeline for ZigBee sensor
technology’. IEEE Computer Magazine, 38, pp. 16–
19.
Hansen, T.R., et al., 2005. ‘Using smart sensors and a
camera phone to detect and verify the fall of elderly
persons’. Proc. European Medical & Biological
Engineering Conference (EMBEC 2005), November
2005.
Nait-Charif, H., and McKenna, S. J., 2004. ‘Activity
Summarisation and Fall Detection in a Supportive
Home Environment’. Proc. of the 17th international
Conference on Pattern Recognition, (Icpr'04),
Washington DC, August 2004, 4, pp. 323–326.
Neurosolution web.
http://www.nd.com/definitions/mlp.htm last visited 2007.
Noury, N., 2002. ‘A Smart Sensor for the remote follow
up of activity and fall detection of the elderly’. Proc.
2nd Annual International Special Topic Conference on
Microtechnologies in Medicine & Biology, IEEE-
MMB2002, Madison-USA, May 2002, pp. 314–317.
Rodriguez J.M., et al, 2005. ‘Portable System for Patient
Monitoring With Wireless Technologies’. The
European Journal for the Information Professional,
October 2005, 5, (5), pp. 46–52.
Sixsmith, A., and Johnson, N., 2004. ‘A smart sensor to
detect the falls of the elderly’, IEEE Pervasive
Computing, April-June 2004, 3, (2), pp. 42–47.
Tunstall web. http://www.tunstall.co.uk last visited 2007.
Williams, G., et al., 1998. ‘A smart fall and activity
monitor for telecare applications’. Proc. of the 20th
IEEE Annual International Conference of the
Engineering in Medicine and Biology Society,
October 1998, 3, pp.1151–1154.
ZigBee Alliance: ‘ZigBee Standard’, 2007
FALL DETECTOR BASED ON NEURAL NETWORKS
545