Assistive Home Platforms

From Guidelines to Technology Selection and Reasoning Applications

Laura Montanini

Department of Information Engineering, Universit

a Politecnica delle Marche, Via Brecce Bianche 12, Ancona, Italy

1 STAGE OF THE RESEARCH

The assistive home technology described in this pa-

per is already available as a Proof of Concept (PoC).

Some aspects, such as, infrastructures and data acqui-

sition are more consolidate; other ones, should be im-

proved and more widely investigated, implementing,

for example, machine learning algorithms for daily

activity recognition. At this development stage, the

research is focused on testing the PoC, evaluating us-

ability of user interfaces and checking false positive

or missed alarms by reasoning algorithms already im-

plemented. The testing phase exploits both simula-

tions and real users. Such users do not ﬁt the target

user, since they are typically young volunteers, but

their contribution could be very useful to improve al-

gorithms and interfaces, and to obtain a feedback on

the system reliability and effectiveness. Based on this

PoC, more reﬁned reasoning algorithms will be im-

plemented. Further improvements foresee the addi-

tion of other sensors, to recognize more precisely the

behavioural activities performed by the user.

2 OUTLINE OF OBJECTIVES

The project presented in this paper aims to introduce

an integrated platform for telecare and Ambient As-

sisted Living (ADL) aimed at prolonging independent

living of the elderly at home. The working hypothe-

sis is to consider a person who normally lives alone.

In this scenario the project presented in this doctoral

contribution refers to the design and development of

a touch screen interface tailored to the needs of the

elderly person, in order to control the home automa-

tion system, monitor the status of some sensors and

help the user in the management of drug therapy. For

this purpose, the criteria of usability and acceptability

of technologies, and more speciﬁcally of touch screen

technologies, by elderly subjects have been widely in-

vestigated.

Another aspect on which the paper is focused is rep-

resented by the extraction and processing of the data

acquired from environmental heterogeneous sensors

to implement server side services necessary for the

proper functioning of the user interface. From these

data it is also possible to obtain quite detailed infor-

mation on the user behaviour. Such information al-

lows us to recognize his daily activities and monitor

the state of health in the long term, noting for example

the vitality level, the sleep duration and the frequency

of eating. For this purpose, some simple algorithms

for presence/absence and wakefulness/sleep recogni-

tion have been implemented.

3 RESEARCH PROBLEM

Numerous studies have shown that in developed coun-

tries the number of elderly people has greatly in-

creased in recent years: elderly people (aged over

65 years) are estimated to rise by 2050 to 19.3%

worldwide (Gavrilov and Heuveline, 2003) . Causes

can be traced in many different factors and, partic-

ularly, great relevance has the medical and scientiﬁc

progress, and the development of public health, which

allows to extend the people’s lives, even when suf-

fering from injuries or diseases. As a result of the

increase in life expectancy, however, the need to cre-

ate a protected and safe environment that allows the

elderly people to best use their residual motor and

cognitive skills is becoming gradually more obvious.

In fact, in (Organisation mondiale de la sant

e et al.,

2004) the authors show that the problem of falls is

a critical issue in the lives of elderly people. Typi-

cally, as a result of a fall, physical activity is reduced

for both the need of healing from injuries, and for

the fear of falling again, leading to muscle atrophy,

less social interactions and, in general, to a reduced

quality of life. Nevertheless, the elderly should be

encouraged to leverage their residual capabilities, to

take care of themselves as much as possible and to

preserve their own autonomy. The day-to-day moni-

toring of daily activities can be extremely useful for

identifying problems, worsening or improvements in

their health state, abnormal behavior or dangerous sit-

Montanini L..

Assistive Home Platforms - From Guidelines to Technology Selection and Reasoning Applications.

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

uation. This kind of monitoring, however, is often

performed with wearable devices, quite intrusive and

affecting the daily habits of the user. These solutions

do not represent a real answer to the problem, as they

require an action by the user that should remember

to wear the sensor every morning. For this reason,

it is necessary to use devices that do not interfere

with the user’s normal activities and are “transpar-

ent” for him. Intelligent environmental technology,

which uses sensors and actuators to dynamically con-

trol the environment of the domestic life, has strongly

grown in recent years: intelligent systems are appear-

ing in nursing homes and assisted living environments

to help caregivers to offer the elderly more comfort

and safety. However, although many older people see

the beneﬁts of environmental technology, they have

never used a computer and have trouble in learning

new technologies; furthermore, many of them do not

like that someone can monitor or control their home

environment. The challenge is therefore to develop

systems, devices and interfaces that older people can

use easily, intuitively and independently in their home

environment, without having to face the difﬁculties of

learning the smart home systems and without feeling

constantly observed.

4 STATE OF THE ART

The expression “Smart Home” is used to indicate

dwellings equipped with devices and technologies

promoting the safety, independence and improving

the user’s quality of life. These technologies allow

to obtain a variety of information about the behaviour

and health of its inhabitants. Most smart homes use

such knowledge to automate the environment and to

provide more comfort, but such technologies can also

be exploited to promote and facilitate the independent

living of vulnerable and elderly subjects. As well de-

scribed by (Rashidi and Mihailidis, 2013), there are

many projects in the literature aimed at creating “as-

sisted” home environment. However, as written by

W. Kearns and L. Normie in Gerontechnology vol.6,

no.3, “The reluctance of older adults and policy mak-

ers to adopt technological change may be described

by the ancient proverb: Better a known devil than an

unknown god”. Therefore, there are two main chal-

lenges to consider during the implementation and de-

sign phase of a smart home for AAL: the choice of

appropriate interfaces and the use of less intrusive as

possible technologies.

4.1 User Interface Design for Elderly

When designing technological solutions addressed to

elderly or disabled people, as it happens in Ambient

Assisted Living (AAL) frameworks, it is very impor-

tant to select human-system interfaces that can match

the user’s abilities, and make the user feel secure and

at ease (Wood et al., 2005). In the design phase of the

user interface, taking into account the usability related

aspects is crucial: if these aspects have been consid-

ered, there are more chances the interface is accepted

positively by users. Consequently, the primary goal is

the ease of use.

Jakob Nielsen (Nielsen, 1995) has provided a list of

10 “heuristic” principles for the interaction design.

Such a list does not refer to a speciﬁc type of inter-

face, but is quite general and establishes guidelines

that must be followed in the design phase in order to

provide a tool that is as much easy to use as possible,

and has a good impact on the user from the begin-

ning. Referring more speciﬁcally to older users, it is

important to note that different age-related disabilities

affect the way elderly people can interact with tech-

nological devices. Chaparro et al. (Chaparro et al.,

1999) and Wood et al. (Wood et al., 2005) have shown

that the use of computer equipped with keyboard and

mouse can cause problems with older people. The

touch interaction is commonly seen as an easier ap-

proach to enable older people to interact with a ma-

chine, because it is a more direct interaction that does

not require an advanced mental model. The most ob-

vious advantage of using the touch screen is, accord-

ing to Greenstein and Arnaud (Helander, 2014), that

the input device is the output device too. Further-

more, touch interaction does not require special mo-

tor skills (Wood et al., 2005), however, the interface

design needs a particular attention, especially when

dealing with the elderly. Among the factors that may

inﬂuence the effectiveness and quality of touch inter-

action it is worth mentioning the hand used (if the

“preferred” or the other one), if the device is used

in static conditions or moving (eg. sitting or walk-

ing ), and the position of the target area (ie the area

which, when clicked or interacted, determines a com-

mand activation) with respect to each hand. The size

of the target area, typically a button, has a great in-

ﬂuence on the accuracy of the interaction, no matter

which hand is used. Park et al. (Park et al., 2008)

have shown that the size of the graphic button have

a signiﬁcant inﬂuence on the number of errors, suc-

cess rate and the optimal pressure; that is, applying

a normal pressure, the larger the size, the lower the

error rate and the higher the rate of success. The re-

sults of a study by Pari et al. (Parhi et al., 2006) have

AssistiveHomePlatforms-FromGuidelinestoTechnologySelectionandReasoningApplications

shown the buttons should be approximately 9.2 mm

wide for a mobile device: with these dimensions, the

target areas are as small as possible, without decreas-

ing performance. Moreover, as Yang (Yang, ) states,

complex control techniques should be avoided - older

people may ﬁnd it difﬁcult to perform sliding and re-

volving gestures on touch screen. While evaluating

the use of web-based interfaces by the elderly, Kur-

niawan and Zaphiris (Kurniawan and Zaphiris, 2005)

found that they prefer the presence of graphics if it is

logically associated with the content, and not only for

decorative purposes. Animated elements tend to con-

fuse older users and therefore should be avoided in

general, while animated avatars are considered use-

ful. The icons should be simple and meaningful, and

large enough to be identiﬁed by people with impaired

vision (Maguire, 1999). All these studies help us to

create an overall picture about the general rules for de-

signing usable touch screen interfaces, which require

a minimal effort in understanding by older users.

4.2 Behavioural Analysis Through

Environmental Sensors

In the literature there are a multitude of AAL systems

aimed at monitoring the daily activities to support in-

dependent living, evaluating in a natural and continu-

ous way the health and cognitive state, providing au-

tomated assistance, and reducing the pressure on rel-

atives or caregivers. In this context, one of the most

important applications is represented by the fall detec-

tion. As previously mentioned, in fact, falls are one of

the main threats to independent life of the elderly sub-

jects. Several systems have been developed for this

purpose, especially based on computer vision, such as

(Gasparrini et al., 2014), or wearable sensors (Huynh

et al., 2014; Pierleoni et al., 2014). Other solutions

focus, instead, on the individual activity detection,

such as moving (walking, standing, etc.), as shown

in (De Santis et al., 2014). In (Zhang et al., 2013) a

platform, called ENLIVEN,using environmental non-

intrusive sensors, has been presented: the system is

able to understand activities and vital signs, and, us-

ing such data, to make decisions. Algorithms based

on fuzzy logic have been used in (Medjahed et al.,

2009) for the recognition of different activities, such

as Sleeping, Getting up, Toileting, etc. In this case,

the monitoring system uses three main subsystems:

two microphones to monitor the sound, a wearable de-

vice that measures physiological data, a set of infrared

sensors for detection of the presence and posture of

the subject. Even in (Fleury et al., 2010), the mon-

itoring system consists of presence sensors (IR), mi-

crophones for the sound and speech recognition and

a wearable sensor; contacts on doors and refrigerator,

temperature sensors and humidity are also added. The

technique used for the activity classiﬁcation exploits

the Support Vector Machines (SVMs). The method

presented in (Nam et al., 2011), based on (Kim et al.,

2009), allows to obtain, instead, data such as time of

occurrence, probability of occurrence, daily time in-

tervals and the relations between the various activity

sets, shown according to a speciﬁc graphical model.

Algorithms that can distinguish patterns of activity

occurring frequently, through the analysis of the tem-

poral relations in a multi-user environment, have been

described in (Jakkula et al., 2007). These algorithms

have been used in the context of the CASAS project

(Rashidi and Cook, 2009), a smart environment pro-

viding a non-invasive assistive tool for dementia pa-

tients at home.

5 METHODOLOGY

The following section describes the platform adopted,

the design choices and the methodologies used in the

development phase. A brief overview on the AAL

platform underlying the project is presented ﬁrst, then

the user interface based on touch screen technology is

explained in detail, paying close attention to the de-

sign choices. Finally, algorithms for the presence and

activity recognition are exposed.

5.1 Context: A Smart Home in AAL

Before discussing the design methodologies used, a

quick overview of the underlying AAL framework

will be provided. Such a framework covers several

aspects of the home living, such as independent liv-

ing, home security, health monitoring and environ-

mental control. From a general point of view, in the

system architecture the information is generated by a

multiplicity of subsystems. They implement speciﬁc

functionalities and, through well deﬁned policies and

rules, send the data to a local server (some speciﬁc

information are delivered remotely), which correlates

all the received data, collects information on the status

of the system and on the user’s habits and behaviour

(see Fig 1).

The system is able to:

• analyse behavioural data in a unobtrusive way on

the long term in order to make a preliminary diag-

nosis compared to the observed changes in habits;

• allow the user to interact with the home environ-

ment, facilitating certain tasks, such as opening or

closing windows and blinds, and turning on or off

ICT4AgeingWell2015-DoctoralConsortium

Figure 1: Different subsystems in the platform architecture.

lights, through the home automation sensor net-

work;

• monitor the power consumption of the house-

hold appliances in order to optimally manage the

spending for the electrical energy;

• collect medical data, such as weight or blood

pressure, acquired through electrical devices and

transmit the data to a health professional or a doc-

tor for diagnosis.

Each subsystem uses different devices, transmission

media (CAN bus, ETHERNET, WI-FI and wireless

sub GHz) and communication protocols. All the ac-

quired data are sent via proper network interfaces to a

central server, to store and process information. The

server also listens for requests from the user interface

to provide adequate information and services.

5.2 User Interface Design and

Prototyping

The user interface relies on three different types of

touch screen devices: tablets, smartphones and ﬁxed

touch screens. Each device has features and offers

different advantages, for this reason it is important to

choose the right device according to which user is in-

terfacing the system. Smart phones are characterized

by a great portability, but due to the small size of the

screen, they do not ﬁt with users affected by visual

impairments or poor motor skills. Conversely, ﬁxed

touch screens allow to display contents even to visual

impaired subjects, but cannot be moved from a room

to another easily. In this sense, the tablet is a good

compromise, ensuring mobility and acceptable screen

size. According with the general rules of interface de-

sign, previously mentioned, some speciﬁc guidelines

for the touch screen interaction have been arranged.

In particular, they relate to four key aspects:

• The target elements design: buttons at least 9.2

mm wide for smartphones, and larger for tablets

and ﬁxed screens will be used. Once the target

is captured, in order to communicate to the user

the success of the operation, a visual or audible

feedback is performed.

• The graphic elements design: among the elements

that inﬂuence the effectiveness and the quality of

interaction, particularly important are the graph-

ics. To facilitate the understanding of the content,

each button has an icon and a text label that spec-

iﬁes its meaning. The used graphics is simple and

intuitive, there are no animations.

• Navigation: while browsing it is important that

the user is always able to know where he or she

is; For this reason, each page is characterized by

a title that deﬁnes its content. Moreover an extra

help provides aid during the navigation. One as-

pect to consider is the navigation hierarchy: it is

not recommended to use deep navigation hierar-

chies that may cause confusion and disorientation

to the user.

• The design of the contents layout: having small

screens, the use of the text has been limited to

the minimum, preferring to use keywords instead

of long sentences. The information and graphics

are concentrated in the central area of the touch

screen; the background and the contents are col-

ored appropriately.

In addition to these criteria, an initial check on the

network connection and popup windows notifying the

results of the performed procedures have been imple-

mented. Moreover, the application ﬁnal appearance

differs automatically according to the screen size, i.e.

the interface is optimized for the used device.

The prototype application uses the Android operat-

ing system, from version 3.0 Honeycomb. It allows

to check the status of the lights and turn them on

or off , activate scenarios and open or close blinds

and windows. Through a speciﬁc service, the user

can obtain the values of the measurements obtained

from electromedical devices and visualize them on

the screen. Monitoring functionalities are also avail-

able for checking the status of the environmental sen-

sors, of the power consumption and of the activities

carried out by the user; an example of graph that sum-

marize the user’s daily or weekly behaviours are vis-

ible in Fig. 2 Finally, additional functionalities allow

to display notiﬁcations to remind to the elder to take

medicines at the prescribed time. All these services

are requested from the local server, according to the

classic client-server model, which answers to each re-

quest using a string formatted in JSON standard.

AssistiveHomePlatforms-FromGuidelinestoTechnologySelectionandReasoningApplications

Figure 2: An example of daily activity graph: a time line

representation of sleeping activity displayed by the user in-

terface.

5.3 Behavioural Analysis

As already widely discussed, the smart home for AAL

allows to monitor and assist the daily activities of the

elder living alone at home, in order to obtain infor-

mation on his or her behaviour. From the analysis of

these information, it is possible to deduce any pro-

gressive worsening in the state of health, potential

risks and emergency situations. One of the strengths

of the project described in the previous subsection is

the ability to collect a large number of behavioural

data without the need for any user action. As already

mentioned, this is an essential aspect in the design of

systems intended for older users. In order to make

the system non-intrusive, environmental sensors have

been used, such as:

• PIR;

• magnetic sensors for fridge, doors and windows;

• pressure sensors for beds and chairs;

• power meters for electrical consumption;

• ﬂow meters for water and gas.

The fusion of the data obtained from these sensors al-

lows us to determine the activities carried out by the

user. The behavioural data that can be identiﬁed are

mainly of three types: energy consumption, presence

and activities. While the ﬁrst concerns only the en-

ergy consumption and thus may be derived directly

from the observation of the data acquired from the

power meter, without using more challenging algo-

rithms , the last two types exploit more complex meth-

ods of data analysis.

5.3.1 Presence Recognition

The presence recognition at home is a very important

point that allows to identify any abnormal situations

or alert. For example the elderly could turn on the

stove and, forgetting it on, leave the home. This rep-

resents a potentially dangerous situation that should

be avoided. To this aim a presence recognition algo-

rithm has been implemented. For the assessment of

the state of presence, only some sensors have been

considered: PIR, magnetic and pressure sensors. The

ﬂow meter and power meter have not been considered

at present. The position of some sensors is essential

for the proper functioning of the algorithm. In par-

ticular it is necessary to monitor the front door (the

working hypothesis is that there is only an entry). For

this purpose two PIRs, one inside and one outside

the door, and also a magnetic sensor which can de-

tect its opening/closing, have been positioned. The

basic concept of the algorithm is that, depending on

the user’s location, the probability that he exited may

be higher or lower; consequently the time needed to

determine if he is absent varies according to this con-

sideration. In order to determine where is the user,

eight states have been identiﬁed:

• Absent;

• User out, door closed: the house is empty and the

user is outside; the door is closed;

• User out, door open: the house is empty and the

user is outside; the door is open;

• User in, door open: the user is in the house, and

the door is open;

• User in, door closed: the user is in the house, and

the door is closed;

• Sleeping/resting: the user is in the house, and he is

sleeping (night time) or resting (during the day);

• Present (generic): the user is in the house, indoor

or outdoor, but we have no more information on

what he is doing or where he is;

• Multiple users: the user is in the house, with

someone else.

The algorithm consists of a state machine in which

events allowing to move from a state to another are

those detected by the sensors (such as opening a door

or activating a PIR sensor).

When a sensor detects an event, the presence relia-

bility is increased by certain percentage value. As

time passes, without the occurrence of other events,

this value progressively decreases. When reliability

reaches zero, the status changes to Absent. The

value of decrease depends on the current state. For

example, if the user is outdoor and the door is closed,

just a few minutes without the occurrence of another

event are sufﬁcient to assert that the user is absent;

while, when the user is in the house and the door

is closed, much more time is necessary before the

system can notify the absence. When multiple events

occur at the same time in different rooms it is possible

to state that there are multiple users. Also in this case

ICT4AgeingWell2015-DoctoralConsortium

Absent

users

Outdoor

door

closed

Outdoor

door

open

Indoor

door

open

Indoor

door

closed

Sleepin

g or

Resting

Present

time

out

time

out

time

out

time

out

time

out

time

out

time out,

no concurrent

event,

bed on

any event

different

from

bed off

indoor event,

closed door

indoor event,

open door

indoor event,

closed door

indoor event,

open door

outdoor event,

closed door

outdoor event,

open door

outdoor

event,

closed

door

indoor event

door open

door open or

door close

bed off,

open

door

bed off,

closed door

bed on

PIR in the

entrance

outdoor

event

outdoor

event

outdoor

event

any event

time out,

no concurrent

event,

bed off

concurrent

event

concurrent

event

outdoor

event

Figure 3: The state diagram of presence recognition algorithm.

when the timer expires, if no further concurrent event

occurs, the system assumes there is only one user.

The state diagram of the algorithm for the presence

recognition is shown in Fig. 3. As obvious, the more

reﬁned is the algorithm, the bigger the state diagram

becomes.

5.3.2 Activity Recognition

Through environmental sensors, it is possible to rec-

ognize the activities carried out by the user. In this

regard, ﬁrst elementary algorithms have been imple-

mented to understand if the user is cooking, sitting

or sleeping. Such information can be obtained using

respectively, a ﬂuxometer on the cooker to detect the

gas ﬂow, a pressure sensor under the mattress of the

bed and a pressure sensor under the cushion of the

sofa. Currently, combining the sensor activations and

deactivations, it is possible to recognize the individual

activity. However, more complex algorithms based

on the sensor fusion, are foreseen to detect activities

more carefully. About that, for example, the activ-

ity of cooking could be recognized by analysing the

data from other sensors, such as those indicating the

use of microwaves, by means of the power meters or

those indicating the opening of the refrigerator, using

magnetic sensors. One of the ideas currently under

study to understand if the user is sitting or is sleeping

consists in implementing truth tables considering the

status of multiple sensors, such as PIR and magnetic

sensors.

6 EXPECTED OUTCOME

As previously explained in section 1, the platform is

already available as a PoC. The system is able to ac-

quire data from sensors and send such information to

the local server in real time.

A ﬁrst prototype of the user interface is already avail-

able. Some screenshots of the application are shown

in Figs. 4 and 5. Respecting the criteria mentioned

above, graphics chosen are very simple and easily un-

derstandable, there are no animations and the buttons

are large sufﬁciently. Moreover different layouts have

been chosen to best ﬁt the size of the screen as you

can see in Figs. 6 and 7. Once the development phase

has been concluded, a veriﬁcation of functions, us-

AssistiveHomePlatforms-FromGuidelinestoTechnologySelectionandReasoningApplications

Figure 4: Screenshot of the user interface that monitors the

current power consumption of different loads (from left to

right: tv, generic load, microwave, generic load and in the

bottom the general meter).

Figure 5: Screenshots of the user interface that controls the

opening/closing of windows and shutters.

ability, reliability and effectiveness of the interface

prototype for user-system interaction has been real-

ized. The system works correctly in all the different

conditions of use. Moreover, in order to obtain an

opinion on usability, some volunteers have tested the

application and ﬁlled out an evaluation form, provid-

ing an overall judgement on the interface. It has been

also request a personal opinion on the level of usabil-

ity of the application when used by elderly people.

The group of volunteers is composed of 13 subjects,

whose average age is 27 years; in Tables 1 and 2 the

evaluations expressed by them are shown.

Although the tests were not carried out by individu-

als with characteristics similar to those of the target

users, they still provide an initial positive response

from people external to the project, considering the

ﬁrst impressions and getting tips for the improve-

Figure 6: Screenshot of the user interface that monitors the

state of some sensors: the application can automatically un-

derstand the size of the screen and choose the right layout

(layout for smartphones).

Figure 7: Screenshot of the user interface that monitors the

state of some sensors: the application can automatically un-

derstand the size of the screen and choose the right layout

(layout for tablets and ﬁxed screens).

ments. Among the various features provided by the

touchscreen application, several concern the visual-

ization of daily or weekly charts of the activities per-

formed by the user. Although more complex or re-

ﬁned algorithms have not been implemented yet, such

graphics allow us to determine the actions with a cer-

tain degree of reliability. The presence recognition

algorithm is currently in the testing phase. From tests

carried out so far, it is able to correctly recognize the

user’s status. Some of the hypothesis made in the de-

sign phase have been changed as a result of tests done

in the laboratory environment, to better adapt to the

real functioning of the system and make it more efﬁ-

cient. The results of the algorithms for the presence

and activity recognition are stored in a database and

ICT4AgeingWell2015-DoctoralConsortium

Table 1: Different aspects evaluated by volunteer users vot-

ing in a range from 1 to 10 (1: very bad, 10: very well.)

Evaluated aspect Rate on the prototype

Visual Appearance:

clarity, visual impact,

coherence between the

different sections

8.7

Intuitive use: easy to

learn how to interact

with the application

8.4

Immediate detection of

functionalities, ease of

navigation

8.8

Efﬁciency: level of

user control, reachabil-

ity of the objectives

8.7

Feedback, notiﬁca-

tions, error handling

7.6

Table 2: Findings on the use within the AAL evaluated by

volunteer users, voting in a range from 1 to 10 (1: very bad,

10: very well).

Evaluated aspect Rate on the prototype

Usability by disabled

or elderly users

7.2

Acceptability and im-

pact on users

7.5

periodically sent to a remote telecare platform which

deals with charts. In Figs. 8 and 9 you can see some

graphs of the data collected in the testing phase.

Figure 8: Screenshot of the remote platform interface show-

ing presence graphs: yellow and orange portions represent

respectively the absence and presence percentage time.

Further improvements foresee the implementation of

more reﬁned algorithm for the activity recognition

and machine learning algorithms in order to predict

the elder’s behaviour, relying on observations of be-

havioural data in the long period.

Figure 9: Screenshot of the remote platform interface show-

ing presence graphs: each color represents the average time

spent for the respective activity (blue for sleeping, green for

cooking, magenta for sitting, grey for undeﬁned).

REFERENCES

Chaparro, A., Bohan, M., Fernandez, J., Choi, S. D., and

Kattel, B. (1999). The impact of age on computer

input device use:: Psychophysical and physiologi-

cal measures. International Journal of Industrial Er-

gonomics, 24(5):503–513.

De Santis, A., Gambi, E., Montanini, L., Raffaeli, L., Spin-

sante, S., and Rascioni, G. (2014). A simple object

for elderly vitality monitoring: The smart insole. In

Mechatronic and Embedded Systems and Applications

(MESA), 2014 IEEE/ASME 10th International Con-

ference on, pages 1–6.

Fleury, A., Vacher, M., and Noury, N. (2010). Svm-based

multimodal classiﬁcation of activities of daily liv-

ing in health smart homes: sensors, algorithms, and

ﬁrst experimental results. Information Technology in

Biomedicine, IEEE Transactions on, 14(2):274–283.

Gasparrini, S., Cippitelli, E., Spinsante, S., and Gambi, E.

(2014). A depth-based fall detection system using a

kinect sensor. Sensors, 14(2):2756–2775.

Gavrilov, L. A. and Heuveline, P. (2003). Aging of popula-

tion. The encyclopedia of population, 1:32–37.

Helander, M. G. (2014). Handbook of human-computer in-

teraction. Elsevier.

Huynh, Q. T., Nguyen, U. D., Tran, S. V., Nabili, A., and

Tran, B. Q. (2014). Fall detection system using com-

bination accelerometer and gyroscope.

Jakkula, V. R., Crandall, A. S., and Cook, D. J. (2007).

Knowledge discovery in entity based smart environ-

ment resident data using temporal relation based data

mining. In Data Mining Workshops, 2007. ICDM

Workshops 2007. Seventh IEEE International Confer-

ence on, pages 625–630. IEEE.

Kim, B., Choi, E., Nam, Y., and Cho, W.-D. (2009). Incre-

mental statistical methods for activity daily pattern ex-

traction and user intention inference. JUCT: Journal

of Ubiquitous Convergence Technology, 3(1):27–34.

Kurniawan, S. and Zaphiris, P. (2005). Research-derived

web design guidelines for older people. In Proceed-

ings of the 7th international ACM SIGACCESS con-

ference on Computers and accessibility, pages 129–

135. ACM.

AssistiveHomePlatforms-FromGuidelinestoTechnologySelectionandReasoningApplications

Maguire, M. C. (1999). A review of user-interface de-

sign guidelines for public information kiosk systems.

International Journal of Human-Computer Studies,

50(3):263–286.

Medjahed, H., Istrate, D., Boudy, J., and Dorizzi, B. (2009).

Human activities of daily living recognition using

fuzzy logic for elderly home monitoring. In Fuzzy

Systems, 2009. FUZZ-IEEE 2009. IEEE International

Conference on, pages 2001–2006. IEEE.

Nam, Y., Rho, S., and Lee, S. (2011). Extracting and vi-

sualising human activity patterns of daily living in

a smart home environment. IET communications,

5(17):2434–2442.

Nielsen, J. (1995). 10 usability heuristics for user

interface design. Nielsen Norman Group, Inter-

net: http://www.nngroup.com/articles/ten-usability-

heuristics/.

Organisation mondiale de la sant

e, B. r. d. l., Skelton, D.,

and Todd, C. (2004). What are the main risk factors

for falls amongst older people and what are the most

effective interventions to prevent these falls? WHO

Regional Ofﬁce for Europe.

Parhi, P., Karlson, A. K., and Bederson, B. B. (2006). Target

size study for one-handed thumb use on small touch-

screen devices. In Proceedings of the 8th conference

on Human-computer interaction with mobile devices

and services, pages 203–210. ACM.

Park, Y. S., Han, S. H., Park, J., and Cho, Y. (2008). Touch

key design for target selection on a mobile phone. In

Proceedings of the 10th international conference on

Human computer interaction with mobile devices and

services, pages 423–426. ACM.

Pierleoni, P., Belli, A., Palma, L., Pernini, L., and Valenti,

S. (2014). A versatile ankle-mounted fall detection

device based on attitude heading systems. In Biomed-

ical Circuits and Systems Conference (BioCAS), 2014

IEEE, pages 153–156. IEEE.

Rashidi, P. and Cook, D. J. (2009). Keeping the resident in

the loop: Adapting the smart home to the user. Sys-

tems, Man and Cybernetics, Part A: Systems and Hu-

mans, IEEE Transactions on, 39(5):949–959.

Rashidi, P. and Mihailidis, A. (2013). A survey on ambient-

assisted living tools for older adults. IEEE journal of

biomedical and health informatics, 17(3):579–590.

Wood, E., Willoughby, T., Rushing, A., Bechtel, L., and

Gilbert, J. (2005). Use of computer input devices

by older adults. Journal of Applied Gerontology,

24(5):419–438.

Yang, T. Appropriate user interface for the elderly.

Zhang, Q., Karunanithi, M., Rana, R., and Liu, J. (2013).

Determination of activities of daily living of indepen-

dent living older people using environmentally placed

sensors. In Engineering in Medicine and Biology So-

ciety (EMBC), 2013 35th Annual International Con-

ference of the IEEE, pages 7044–7047. IEEE.

ICT4AgeingWell2015-DoctoralConsortium