FALL DETECTOR BASED ON NEURAL NETWORKS

Rubén Blasco, Roberto Casas, Álvaro Marco, Victorián Coarasa, Yolanda Garrido, Jorge L. Falcó

2008

Abstract

Falls are one of the biggest concerns of elderly people. This paper addresses a fall detection system which uses an accelerometer to collect body accelerations, ZigBee to send relevant data when a fall might have happened and a neural network to recognize fall patterns. This method presents improved performance compared to traditional basic-threshold systems. Main advantage is that fall detection ratio is higher on neural network based systems. Another important issue is the high immunity to events not being falls, but with similar patterns (e.g. sitting in a sofa abruptly), usually confused with real falls. Minimization of these occurrences has big influence on the confidence the user has on the system.

Paper Citation

in Harvard Style

Blasco R., Casas R., Marco Á., Coarasa V., Garrido Y. and L. Falcó J. (2008). FALL DETECTOR BASED ON NEURAL NETWORKS . In Proceedings of the First International Conference on Bio-inspired Systems and Signal Processing - Volume 2: BIOSIGNALS, (BIOSTEC 2008) ISBN 978-989-8111-18-0, pages 540-545. DOI: 10.5220/0001066205400545

in Bibtex Style

@conference{biosignals08,
author={Rubén Blasco and Roberto Casas and Álvaro Marco and Victorián Coarasa and Yolanda Garrido and Jorge L. Falcó},
title={FALL DETECTOR BASED ON NEURAL NETWORKS},
booktitle={Proceedings of the First International Conference on Bio-inspired Systems and Signal Processing - Volume 2: BIOSIGNALS, (BIOSTEC 2008)},
year={2008},
pages={540-545},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0001066205400545},
isbn={978-989-8111-18-0},
}

in EndNote Style

TY - CONF
JO - Proceedings of the First International Conference on Bio-inspired Systems and Signal Processing - Volume 2: BIOSIGNALS, (BIOSTEC 2008)
TI - FALL DETECTOR BASED ON NEURAL NETWORKS
SN - 978-989-8111-18-0
AU - Blasco R.
AU - Casas R.
AU - Marco Á.
AU - Coarasa V.
AU - Garrido Y.
AU - L. Falcó J.
PY - 2008
SP - 540
EP - 545
DO - 10.5220/0001066205400545