ULTech to Observe Elderly’s Behavior Changes over Time in SH
131
acceleration data can thus be analyzed in conjunction with the ECG signals to
detect anomalies in the user’s behavior and heart-related problems.
Another detection strategy was proposed by Sprint et al. [5]. First, sensor
data are labeled to correspond to activity “to sleep”. Features are then extracted
and used as inputs to change detection algorithms such as RuLSIF, virtual classifier, and sw-PCAR to detect and analyze behavior changes that accompany
health events. If the change is significant, change analysis is performed to explain
the source of change. Use cases studied in this context concern older adults who
experienced major health events, including cancer treatment and insomnia.
Anomaly detection systems for detecting abnormal behavior has been surveyed and reviewed in [6–8] implicitly rely of representation of the human activity
in a spatiotemporal context highlighting various techniques/methods (classification, clustering, nearest neighbor, statistical).
Previous works describe existing research regarding the analysis of behavior
and health monitoring from a smart home. A set of these works [1,4] only use
wearable sensors to monitor vital signs. Works in [2,3,5–8] consider home sensors
to monitor daily activities but do not analyze all activities of the elderly person at
the same time. For example, [5] studied the behavior change related to sleeping
only. All the solutions mentioned previously have been developed to quickly
detect and react as soon as possible when a sudden behavior change occurs,
especially “the fall” of the monitored person. Our objective in this work is, not
only to detect sudden changes, but also to analyze the possible evolution of the
behavior over a long period of time.
3 Our Approach
The overall objective of this study is to analyze the daily behavior of elderly
people in their apartment through ambient sensors. In the following, we introduce our model to characterize the normal behavior pattern for elderly people.
This normal behavior pattern can then be used to detect behavior changes over
time by comparing the current behavioral data of an elderly with her/his usual
behavior pattern.
3.1 Activities and Daily Behavior Pattern
Activities of Daily Living (ADLs) is a term used by healthcare professionals
to refer to the basic self-care tasks an individual does on a day-to-day basis. These
fundamental activities are crucial for maintaining independence. They are used
by health professionals as a way of measuring an individual’s functional status,
especially for elderly people.
The importance of this issue has led to the development of numerous solutions
that can monitor activities (e.g., [9]). Basic ADLs are self-care activities routinely
performed which include, but are not limited to seven activities: sleeping, getting
dressed, eating (three times per day), going to the toilet, hygiene activities (to
take shower and/or bathing) and going outside.
131
acceleration data can thus be analyzed in conjunction with the ECG signals to
detect anomalies in the user’s behavior and heart-related problems.
Another detection strategy was proposed by Sprint et al. [5]. First, sensor
data are labeled to correspond to activity “to sleep”. Features are then extracted
and used as inputs to change detection algorithms such as RuLSIF, virtual classifier, and sw-PCAR to detect and analyze behavior changes that accompany
health events. If the change is significant, change analysis is performed to explain
the source of change. Use cases studied in this context concern older adults who
experienced major health events, including cancer treatment and insomnia.
Anomaly detection systems for detecting abnormal behavior has been surveyed and reviewed in [6–8] implicitly rely of representation of the human activity
in a spatiotemporal context highlighting various techniques/methods (classification, clustering, nearest neighbor, statistical).
Previous works describe existing research regarding the analysis of behavior
and health monitoring from a smart home. A set of these works [1,4] only use
wearable sensors to monitor vital signs. Works in [2,3,5–8] consider home sensors
to monitor daily activities but do not analyze all activities of the elderly person at
the same time. For example, [5] studied the behavior change related to sleeping
only. All the solutions mentioned previously have been developed to quickly
detect and react as soon as possible when a sudden behavior change occurs,
especially “the fall” of the monitored person. Our objective in this work is, not
only to detect sudden changes, but also to analyze the possible evolution of the
behavior over a long period of time.
3 Our Approach
The overall objective of this study is to analyze the daily behavior of elderly
people in their apartment through ambient sensors. In the following, we introduce our model to characterize the normal behavior pattern for elderly people.
This normal behavior pattern can then be used to detect behavior changes over
time by comparing the current behavioral data of an elderly with her/his usual
behavior pattern.
3.1 Activities and Daily Behavior Pattern
Activities of Daily Living (ADLs) is a term used by healthcare professionals
to refer to the basic self-care tasks an individual does on a day-to-day basis. These
fundamental activities are crucial for maintaining independence. They are used
by health professionals as a way of measuring an individual’s functional status,
especially for elderly people.
The importance of this issue has led to the development of numerous solutions
that can monitor activities (e.g., [9]). Basic ADLs are self-care activities routinely
performed which include, but are not limited to seven activities: sleeping, getting
dressed, eating (three times per day), going to the toilet, hygiene activities (to
take shower and/or bathing) and going outside.