Instrumentation & Measurement Magazine 24-6 - 83

assessing the behavior in a kitchen for example, the frequency
of light switch use, or toaster use in food preparation. A stove
top thermal infrared camera in addition to a rule-based algorithm
has been used to alert users of potential danger such as a
fire [59], [60]. Microphones have been used to detect user activities
such as eating, drinking, talking, laughing and coughing
[61]-[63]. Wi-Fi channel state information (CSI) has been used
to detect gestures and biosignals such as respiration in addition
to being able to identify people [61]. A recent study used a
Wi-Fi system called WiFall that utilizes CSI to detect a fall in an
indoor environment [61].
Ambient Sensing Outside the House
Inherent in the nature of ambient sensing; once the subject
leaves the house, they are no longer being monitored by
the home-based sensors. Ambient sensors do not necessarily
need to be embedded in the home and can be embedded
in an individual's car to collect, monitor and assess driving
behavior. Driving skill decline can be identified by such variables
as repeated route selection and backtracking strategies.
GPS has been used to track cognitive change over time in relation
to baseline values [64]. Accelerometric data was used in
the evaluation of a control group and frailty impaired drivers
to identify unsafe driving behaviors whether due to cognitive,
physical or general health declines [65], [66]. Driving complexity
has been evaluated among test subjects using GPS
data to determine cognitive discriminative capacity in a study
on naturalistic driving behavior. The findings reveal reduced
complexity, frequency and distance in line with cognitive declines
[67], [68].
A common presentation of advancing dementia is nighttime
wandering, and one that requires ambient sensing
capable of detecting such an anomaly [69], [70]. Wandering
patterns can be categorized into four types: direct, lapping,
random and pacing [71]. Ambient sensors can aid in two main
purposes; first the detection of wandering and second, the deterrence
of wandering [69], [70]. Detection of wandering can be
accomplished with many sensor types; a 12-week study was
carried out on 5 subjects in residences equipped with a monitoring
system [69]. The monitoring sensors included motion
sensors, bed sensors, floor sensors, door sensors, window sensors
and latch sensors, and the intervention devices included a
speaker and smart light [69]. The results from this study indicated
that the system was successful in identifying wandering
behavior, helpful in re-directing the subject from their wandering
behavior back to bed, and resulted in increased sleep
quality of the dementia subject's caregiver [69].
Cognitive Training
There is much research focused on not only monitoring cognitive
decline but on identifying and improving this decline
through cognitive training [72], [73]. Studies have suggested
that the brain is capable of learning throughout it's lifetime,
where computer programs aiming to facilitate cognitive training
have led to improved cognitive functioning [73]. As a
result, there has been significant interest in the development
September 2021
of cognitive training games. These games offer the ability to
assess brain function as well as motor skills through specific
measures. For example, a computerized Sudoku game and a
word search game were developed in [74] for subjects suffering
from mild cognitive impairment (MCI). The Sudoku game
measured mouse movements, solution entry and deletions,
pencil mark entry and deletions and number of hints provided.
The word search game measured mouse movements,
letter selection and de-selection, number of words found and
number of hints provided [74]. These measurements were
used to evaluate the performance of MCI subjects in game
play, where the results suggested that these games could differentiate
between cognitive performances and would provide
long term insights into cognitive behavior and trajectory. Informed
by these results, the same group developed another
game based on 'whack-a-mole' that was designed for subjects
with moderate dementia, and administered over a one-year
time frame [72]. Twelve subjects participated in the study and
demonstrated proficient use of a tablet for gaming purposes in
addition to a remarkable ability to improve (procedural learning),
despite experiencing cognitive decline. Trajectories in
individual game performances were directly correlated with
clinical cognitive assessment trajectories, indicating that the
use of this game in evaluating the cognitive status of moderate
dementia patients may be effective and even predictive of adverse
outcomes [72].
Summary
One of the challenges in summarizing ambient sensing of
ADLs is the diversity in sensor types and sensor systems. The
number of sensor types used in ambient sensing of activity is
as numerous as the permutations of sensor combinations in
these sensor systems [75]. In a 2019 systematic review of technologies
for monitoring lifestyle habits (ADLs) related to brain
health, only 15 papers met the inclusion criteria [75]. Not only
are there innumerable sensor systems, but often the terminology
is highly specific, making it difficult to account for all
available ADL monitoring sensor systems. Of the 15 studies
identified, most of the studies were multimodal in nature and
at most three studies used the same main technology to monitor
the subjects [75]. This paper reviews some sensor systems
used for monitoring ADLs but is certainly not an exhaustive
review. Table 2 lists the sensor systems and applications discussed
in ambient sensing of ADLs.
Time Dimension in Measuring
Movement and Activity
In examining movement and activity, time is a necessary
dimension. In the previous sections, movement and activity
were described in relation to ambient sensing, where the
intricacies of examining multiple types of sensor data longitudinally
were not discussed. The examination of movement and
activity on a longitudinal basis can be considered analogous
to the examination of human behavior, which can be understood
as patterns in a sequence of movement and activity
observations [15]. Fig. 2 illustrates the interplay between time,
IEEE Instrumentation & Measurement Magazine
83
Instrumentation & Measurement Magazine 24-6

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 24-6
