Page 151 - Petelin, Ana, ur. 2021. Zdravje starostnikov / Health of the Elderly. Zbornik povzetkov z recenzijo / Book of Abstracts. Koper: Založba Univerze na Primorskem/University of Primorska Press
P. 151
vacy preserving fall sensing tehnološke in ostale ikt rešitve | technological an ict solutions
Niki Hrovatin1,2, Aleksandar Tošić1,2, Jernej Vičič2
1 InnoRenew CoE, Izola, Slovenia
2 University of Primorska, Faculty of Mathematics, Natural Sciences
and Information Technologies, Koper, Slovenia
Introduction: Falls are unpredictable accidental events. Common in childhood,
rare in adulthood, but a significant problem among the elderly. Although sever-
al factors reinforce fall prevention, the fall could be an inevitable effect of oth-
er health related complications. Therefore, immediate identification of a fall
event could prevent severe consequences of fall related injuries and other dan-
gerous episodes. Fall detection is commonly addressed using technology de-
signed to differentiate a fall event from activities of daily living. However, tech-
nology adoption by older adults is lower than in other demographic segments,
and fall detection systems based on wearable sensors require active user in-
teraction through battery charging, wearing, and maintenance. Therefore pas-
sive solutions that do not require active user interaction should encounter few-
er barriers to adoption. We present the recent development of a fall detection
system embedded in a composite floor designed to be easy to install, modu-
lar, and cost effective. Moreover, we plan to take advantage of the peculiarity
of sensing only the force applied to the floor, to develop a privacy aware solu-
tion, that guarantees privacy, and gives data ownership solely to the end user.
Methods: Our research is centered on a pilot implementation named smart
floor, a floor system measuring 120cm×120cm, with a standard laminate floor-
ing surface over a layer of rolled foam insulating subfloor. Below this layer, 16
force sensors are placed and wired to an Arduino micro-controller. The force
applied on the floor, for example, by walking across, is measured and sent to an
external system for data capture and processing. The smart floor was used to
acquire data about simulated falls. The data was collected on a data gathering
event organized in a properly equipped gym. Each participant performed sev-
en different fall events selected from the guidelines defined in a previous study.
Results: Data acquired in the data collection process was grouped in a data-
set consisting of 420 fall event records, each record associated with partici-
pants’ demographic characteristics. The dataset was used to train artificial in-
telligence models to detect fall events occurring on the smart floor.
Discussion and conclusions: The preliminary results are promising, but the smart
floor is quite distant from the end implementation. Current artificial intelli-
gence models were trained with data about simulated falls and not real acci-
dental falls that may be affected by objects within the environment. Moreover,
the collected dataset does not include older persons (+60) fall events, which
may affect the system differently. Undeniably, fall detection is the main value of
the developing system. However, the floor as a non-intrusive monitoring sys-
tem and the privacy approach could be determinant aspects for the adoption
of the proposed smart floor.
Keywords: fall detection, smart floor, privacy, artificial intelligence
149
Niki Hrovatin1,2, Aleksandar Tošić1,2, Jernej Vičič2
1 InnoRenew CoE, Izola, Slovenia
2 University of Primorska, Faculty of Mathematics, Natural Sciences
and Information Technologies, Koper, Slovenia
Introduction: Falls are unpredictable accidental events. Common in childhood,
rare in adulthood, but a significant problem among the elderly. Although sever-
al factors reinforce fall prevention, the fall could be an inevitable effect of oth-
er health related complications. Therefore, immediate identification of a fall
event could prevent severe consequences of fall related injuries and other dan-
gerous episodes. Fall detection is commonly addressed using technology de-
signed to differentiate a fall event from activities of daily living. However, tech-
nology adoption by older adults is lower than in other demographic segments,
and fall detection systems based on wearable sensors require active user in-
teraction through battery charging, wearing, and maintenance. Therefore pas-
sive solutions that do not require active user interaction should encounter few-
er barriers to adoption. We present the recent development of a fall detection
system embedded in a composite floor designed to be easy to install, modu-
lar, and cost effective. Moreover, we plan to take advantage of the peculiarity
of sensing only the force applied to the floor, to develop a privacy aware solu-
tion, that guarantees privacy, and gives data ownership solely to the end user.
Methods: Our research is centered on a pilot implementation named smart
floor, a floor system measuring 120cm×120cm, with a standard laminate floor-
ing surface over a layer of rolled foam insulating subfloor. Below this layer, 16
force sensors are placed and wired to an Arduino micro-controller. The force
applied on the floor, for example, by walking across, is measured and sent to an
external system for data capture and processing. The smart floor was used to
acquire data about simulated falls. The data was collected on a data gathering
event organized in a properly equipped gym. Each participant performed sev-
en different fall events selected from the guidelines defined in a previous study.
Results: Data acquired in the data collection process was grouped in a data-
set consisting of 420 fall event records, each record associated with partici-
pants’ demographic characteristics. The dataset was used to train artificial in-
telligence models to detect fall events occurring on the smart floor.
Discussion and conclusions: The preliminary results are promising, but the smart
floor is quite distant from the end implementation. Current artificial intelli-
gence models were trained with data about simulated falls and not real acci-
dental falls that may be affected by objects within the environment. Moreover,
the collected dataset does not include older persons (+60) fall events, which
may affect the system differently. Undeniably, fall detection is the main value of
the developing system. However, the floor as a non-intrusive monitoring sys-
tem and the privacy approach could be determinant aspects for the adoption
of the proposed smart floor.
Keywords: fall detection, smart floor, privacy, artificial intelligence
149
