A Remote Home Monitoring System to Support Informal Caregivers

of People with Dementia

Stefan Lentelink

, Monique Tabak

1,2

, Boris van Schooten

, Dennis Hofs

Harm op den Akker

and Hermie Hermens

1,2

Telemedicine Group, Roessingh Research and Development, Enschede, The Netherlands

Telemedicine Group, University of Twente, Enschede, The Netherlands

Keywords: Dementia, Informal Caregiver, Aging in Place, eHealth, Health Informatics, Ambient Assisted Living,

Remote Home Monitoring.

Abstract: Informal caregivers of people with dementia have a high risk of becoming overburdened. Health informatics

for aging in place can provide them support by deploying unobtrusive remote home monitoring systems to

assess real-time events and monitor changes in the behavior of the person with dementia (PwD). In this paper,

we describe the concept, development, and evaluation of an intelligent remote Home Monitoring System

(HMS) that provides support to informal caregivers by giving key information related to the health and

independent living of the PwD. The HMS consists of a Sensor System that monitors low-level behaviors of

the PwD, a Decision Support System that translates this into high-level behaviors, and a connected

Smartphone Application that allows the caregiver to receive notifications, review behavioral information at a

glance, and facilitates the collaborative care process between informal caregivers. The final HMS prototype

was evaluated and scored high in terms of usability and quality of the Smartphone Application. The Sensor

System showed no significant flaws during testing, and the Decision Support System is considered a viable

proof of concept. The next step is to evaluate the HMS in a real-life setting in terms of offering peace of mind

and reducing the burden of care.

1 INTRODUCTION

The world population is aging, resulting in an

increasing number of people suffering from dementia.

In the Netherlands, 1 in 5 people develop dementia

(VUmc Alzheimercentrum et al. 2012). In its early

stages, people are able to remain at home with the

support of informal caregivers who provide the

majority of care, and who have a key role in

determining the person’s wellbeing (Ministry of

Health Welfare and Sport et al. 2009). The

dependency on informal caregivers in society is

increasing, many of whom face difficulties in their

caregiving tasks and are (at high risk of becoming)

overburdened (Ministry of Health Welfare and Sport

et al. 2009; VUmc Alzheimercentrum et al. 2012).

Providing them with sufficient support is therefore of

great importance, but also to ensure that those with

dementia can continue to live at home longer.

Technological innovations that promote aging in

place, such as Ambient Assisted Living (AAL), could

provide a solution by implementing unobtrusive

remote home monitoring systems that employ a

network of sensors to assess real-time events and

monitor changes in the behavior of a person. AAL has

the potential to promote the quality of life and safety

for people with dementia, give peace of mind to the

informal caregivers, and promote independence and

autonomy for both target groups (Alzheimer’s

Society 2017). This fits well with informal

caregivers’ need for reassurance in terms of

(Instrumental) Activities of Daily Living ((I)ADL)

and safety in the home of the person with dementia

(Kirsi et al. 2004; Pollitt et al. 1991; Bank et al. 2006;

Nolan et al. 2002).

However, the majority of AAL and related

systems found in systematic reviews do not report on

informal caregivers as end-users (Ienca et al. 2017;

Liu et al. 2016; Carswell et al. 2009). This is also

reflected in studies on caregiver burden (Peeters,

Werkman and A Francke 2014; Peeters, Werkman

and AL Francke 2014; Zwaanswijk et al. 2013;

Miranda-Castillo et al. 2013; Peeters et al. 2010),

Lentelink, S., Tabak, M., Schooten, B., Hofs, D., Akker, H. and Hermens, H.

A Remote Home Monitoring System to Support Informal Caregivers of People with Dementia.

DOI: 10.5220/0006565300940102

In Proceedings of the 11th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2018) - Volume 5: HEALTHINF, pages 94-102

ISBN: 978-989-758-281-3

where informal caregivers are rarely surveyed

about using AAL to support them in care tasks.

Clearly, we can observe that support for informal

caregivers of people with dementia is needed.

As part of the H2020 project IN LIFE

(http://www.inlife-project.eu), we developed and

implemented a Health Monitoring Application, called

HELMA. This monitoring tool aims to inform

informal caregivers about the health and wellbeing of

the person with dementia over the long-term by

means of short frequent online questionnaires. To

improve the support for informal caregivers, and to

make the system more discreet and time-efficient, we

aim to improve HELMA with objective monitoring.

Motion and door sensors are proven to be useful

to quantify (changes in) ADL (Yang and Hsu 2012)

(Urwyler et al. 2017), are readily accepted in society

(Pol et al. 2016), appear to be almost unnoticed by

residents after installation in their homes (Nijhof et al.

2013), and are relatively simple and not too expensive

(Peetoom et al. 2015). In addition, simple estimation

methods can be used to quantify daily rhythms (Yang

& Hsu 2012), and measuring baselines are useful to

find changes in behavior (Glascock and Kutzik 2000).

As such, we have developed a Health Monitoring

System (HMS): a remote home monitoring system to

support informal caregivers of people with dementia

in their caregiving tasks. In this paper, we describe

the concept, development, and evaluation of this

remote home monitoring system.

2 METHODS

The work described in this paper is based on an

iterative, user-centered design approach as shown in

Figure 1. By involving potential end-users in the

design process, we aim to increase the usability and

usefulness of the developed system. The HMS

consists of three main components: (1) Smartphone

Application; (2) Sensor System; and (3) Server.

First, we performed a state-of-the-art study by

searching in among others scientific databases

(e.g., Scopus, PubMed, and ScienceDirect) on the

topics of dementia, informal caregivers, remote home

monitoring, user interface design, and state-of-the-art

of remote home monitoring systems. Based on our

findings we developed a scenario following the

approaches of PACT (People; Activities; Context;

and Technology) (Huis in ‘t Veld et al. 2010) and

FICS (Function and events; Interactions and usability

issues; Content and structure; and Style and

aesthetics) (Benyon and Macaulay 2002). Starting

from the scenario, potential end-users were involved

to collect, elaborate, and refine the HMS

requirements. The end-users were contacted via local

healthcare organizations, and consisted of small

groups based on the assumption that the best results

in terms of usability testing come from no more than

5 end-users and performing as many small tests as

possible (Nielsen and Landauer 1993). During the

evaluations, which were divided into three phases, the

Smartphone Application was central.

Figure 1: General iterative design process of the HMS.

Evaluation Smartphone Application

Phase 3

Literature

Scenario (PACT & FICS)

Functional & technical

requirements

Available systems

Decision Support System

Integration

controller & sensors

Technical evaluation

Sensor System

Server

Phase 1

Semi-structured interview:

(in)formal caregivers (n=3)

Low-fidelity prototype

Semi-structured interview:

(in)formal caregivers (n=3)

Pase 2

Semi-structured interview,

tasks, and SUS:

(in)formal caregivers (n=6)

High-fidelity prototype

Semi-structured interview,

tasks, SUS, and uMARS:

non-caregivers (n=5)

Final prototype

Smartphone Application

Semi-structured interview,

tasks, SUS, and uMARS:

(in)formal caregivers (n=5)

A Remote Home Monitoring System to Support Informal Caregivers of People with Dementia

Smartphone Application: In phase 1, we drafted

the first requirements of the HMS and created a low-

fidelity prototype of the application based on the

scenario. The prototype was evaluated in two

successive sessions using semi-structured interviews

on demographics, HMS functionalities, and the

designed prototype. Each session included two

informal caregivers of people with dementia and one

dementia case manager (total n=6).

In phase 2, a high-fidelity prototype was created

for Android-based smartphones. The prototype was

evaluated using a mixed-methods approach,

including a semi-structured demographics interview,

task-based think-aloud protocol, controlled

observation, System Usability Score (SUS) (Brooke

1996), and a semi-structured interview in terms of

usefulness. The evaluation included the four informal

caregivers and two case managers from phase 1.

In phase 3, a final prototype was created

corresponding to the approach in phase 2. In addition,

the evaluation included two successive sessions and

the User Version of the Mobile Application Rating

Scale (uMARS) (Stoyanov et al. 2016) (using

multiple-forward translation). The first session

included five non-caregivers, and the second session

included three informal caregivers of people with

dementia and two dementia case managers.

The resulting functional requirements of the HMS

were prioritized by the first author (SL).

Sensor System: Based on the requirements, we

searched for a suitable Sensor System based on

current systems in literature and commercially

available systems. The Sensor System should have a

suitable controller with open API, bi-directional

communication, support for motion and door sensors,

and should be commercially available in the

Netherlands. The necessary motion and door sensors

should be simple, small, compatible with the chosen

controller, and commercially available in the

Netherlands.

Server: The Server consists of the Decision

Support System (DSS), which interprets the data

collected by the Sensor System, and the Database

Server. For the DSS, the requirements were used as a

guiding principle to determine how and what should

be monitored.

The technical evaluation of the DSS, and indirect

of the Sensor System, was conducted in a 5-day in-

home test where the first author (SL) annotated all in-

home activities (see Section 7), while being

monitored by the Sensor System (see Figure 5). The

outcome of the annotations was compared to the data

of the DSS to explore the reliability of the system.

Table 1: Functional requirements of the HMS; showing five

high-priority examples.

Functional requirement – HMS

The HMS should:

… support multiple users

… show information about events and behavioral

changes

… differentiate between normal, abnormal, and

alarming situations

… include a shared calendar

… include a chat function

3 HMS CONCEPT

The HMS concept can be described as follows:

The Health Monitoring System includes multiple

wireless sensors that can be easily placed in the home

of the person with dementia (i.e., the resident). All

these sensors together unobtrusively monitor the

activities and behavioral changes of this person.

The informal caregivers will be remotely informed

via a smartphone application about the home

situation, and receive notifications upon meaningful

events and behavioral changes regarding the person

they care for.

Table 1 shows a selection of five high-priority

functional requirements of the HMS (as determined

by the first author (SL), based on functionality,

desirability, and feasibility). The original functional

requirements varied in terms of specificity, ranging

from for example “should support multiple users” to

“should only display the latest event or notification in

the home screen”. The functional requirements with

respect to the parameters to be monitored by the HMS

were divided into Detections (basic activities and

actions) and Patterns (slow changes and unusual

behavior). The Detections are shown in Table 2. The

Patterns are defined as deviations on the Detections,

based on the person’s standard behavior. An example

is: “The HMS should monitor deviations in the time

of going to bed”.

The Smartphone Application should primarily

provide key information (i.e., on location, activity,

sleeping, and eating) related to the health and

independent living of the person with dementia. It is

also important to support and improve communi-

cation between informal caregivers, for example by

including a shared calendar and chat function.

The Sensor System should include multiple

wireless sensors that can be easily placed in the home

of the person with dementia. All these sensors

together should unobtrusively monitor the activities

HEALTHINF 2018 - 11th International Conference on Health Informatics

and behavioral changes of this person. The sensors

should be simple, small, and largely respect privacy.

Finally, the Server should be a secure

environment for the collected data. In addition, the

DSS should be reliable and complete in terms of the

activities and behaviors to be measured.

Table 2: Functional requirements of the HMS Detections.

Functional requirement – HMS Detection

The HMS should inform about location by monitoring:

… leaving the house

… outdoor location

The HMS should inform about sleeping by monitoring:

… time of going to bed

… time of getting out of bed

… sleep duration

… number of times and time out of bed

… restlessness

The HMS should inform about activity by monitoring:

… wandering

… toilet usage

D10

… physical activity

D11

… social activity

The HMS should inform about eating by monitoring:

D12

… meals (breakfast, lunch, and dinner)

D13

… drinking

The HMS should inform about … by monitoring:

D14

… medication intake

D15

… personal hygiene

D16

… body weight

D17

… falling

D18

… appliances

D19

… fire alarm

D20

… in-home temperature

D21

… (unwanted) visitors

4 ARCHITECTURE

The architecture of the HMS consists of: (1) Sensor

System; (2) Server; and (3) Smartphone Application

(see Figure 2).

The Sensor System consists of a controller and

several door/motion sensors. It monitors the home of

the person with dementia (i.e., the resident) and sends

all sensor detections to the DSS.

The DSS receives and stores the sensor detections

received from the Sensor System. Algorithms then

analyze and process the data, and generate events.

The DSS can generate events on multiple levels (see

Section 7), which are written to the Database Server.

The Database Server consists of a MySQL back-

end and secure API that both communicate with the

DSS and the Smartphone Application. It contains the

data of all residents, and informal caregivers, based

on the DSS and Smartphone Application. The DSS

generates sensor events, whereas the Smartphone

Application generates chat messages and shared

calendar items. Every data record is linked to a

resident ID that determines the access control.

Informal caregivers can only access data from their

own resident.

The Smartphone Application consists of the Local

Database, which is automatically synchronized with

the Database Server, and the User Interface,

consisting of all the necessary components for the

user interface interaction. The Local Database is a

replication of part of the Database Server and only

contains relevant data for its user. The Local Database

allows the user to use the Smartphone Application

offline and improves the user experience.

Figure 2: Architecture of the HMS.

Server

Smartphone Application

Sensor System

Decision Support

System

Database Server

Controller

Sensors

User Interface

Local Database

A Remote Home Monitoring System to Support Informal Caregivers of People with Dementia

5 SMARTPHONE APPLICATION

The Smartphone Application consists of three main

components: (1) Home; (2) Calendar; and (3) Chat

(see Figure 3).

The Home screen displays clear information

regarding the person with dementia (i.e., the resident).

There are three states for displaying information:

(1) normal; (2) abnormal; and (3) alarming (see

Figure 4). In the normal state, general information is

displayed about the resident (e.g., “The resident went

to bed.”). In the abnormal state, information is

displayed which does not require immediate

attention, but must be kept an eye on (e.g.,

“The resident was 2 hours out of bed last night.”).

And in the alarming state, immediate action is

required (e.g., “The resident left the house at night!”).

An overview of all notifications is shown in Table 3.

The Home screen is also the gateway to many

other features of the application. Starting at the top,

there are four buttons. In order from left to right, the

Phone button allows the informal caregiver to call the

resident directly, but also to temporarily block the

calls from the resident. This measure is implemented

because some residents call their informal caregivers

extremely often, leading to frustration of the informal

caregiver. When the calls are blocked, the caller (i.e.,

the resident) will be redirected to the voicemail.

Informal caregivers are therefore advised to set a

reassuring voicemail beforehand. The Mute button

allows the informal caregiver to set all HMS

notifications to sound and vibrate, vibrate, or mute.

This does not affect other phone notifications. The

Connection button allows the informal caregiver to

check the connection with the HMS. And the Settings

button allows the informal caregiver to view and

adjust various system settings, such as Account,

Notifications, Calls, Events, and Installation. The

Home screen also contains a Menu button, allowing

the user to view current and historical notifications

about each of the Location, Activity, Sleeping, and

Eating domains (see Figure 4). The user can view all

aforementioned domains in a single view in

Overview. Each domain also gives the user the

opportunity to view data graphically per week,

month, quarter, or half year, with a view to inform

about behavioral changes. The Calendar gives

informal caregivers the opportunity to schedule

mutual appointments. And the Chat gives informal

caregivers the option to send each other messages.

The Smartphone Application follows the general

design principles to clearly show its users at a glance

all the necessary information.

The phase 3 evaluation with the final prototype

showed a median (range) SUS score of 87.50 (15.00),

and a uMARS score of 4.27 (0.73) with the

following subdomains: Engagement 4.00 (1.60);

Functionality 4.25 (0.75); Aesthetics 4.33 (1.00); and

Information 4.50 (1.00).

In addition, the interviews highlighted improve-

ments such as:

• The application should include the option to

assign caregivers to calendar items.

• The application should include the option to

add notes, photos, and contact information to

calendar items.

• The application should geographically

prioritize caregivers in case of alarming

situations, so that geographically close

caregivers are warned first. They should

however be able to forward the alarm.

Figure 3: Smartphone Application; showing the Calendar on the left, the Home screen in the center, and the Chat on the right.

HEALTHINF 2018 - 11th International Conference on Health Informatics

Figure 4: Smartphone Application; showing the Sleeping

overview on the left, the Home screen during an abnormal

event in the center, and the Home screen during an alarming

event on the right.

Table 3: Overview of the notifications implemented in the

Smartphone Application.

Smartphone Application notification

Location notification about:

… coming / leaving home during the day

… coming / leaving home at night

… leaving home for a long period of time

… not coming home at night

Sleeping notification about:

… going to bed

… getting out of bed

… not going to bed

… not getting out of bed

… sleep duration

N10

… number of times and time out of bed

Activity notification about:

N11

… physical in- or overactivity

N12

… social inactivity

N13

… remarkable toilet usage

Eating notification about:

N14

… forgetting / preparing too many meals

(breakfast; lunch; and dinner)

Other notification about:

N15

… in-home temperature

N16

… new calendar item

N17

… new chat message

6 SENSOR SYSTEM

In our search for a controller, thirty products were

found varying from complete systems to standalone

controllers. Of these, the following five products

fitted the requirements: (1) Insteon; (2) Fibaro; (3)

Raspberry Pi + RaZberry + Domoticz; (4) Eedomus;

and (5) Vera. Other systems, such as Samsung

SmartThings, BeNext, Zipato, D-Link, Devolo,

Wink, and others, were unsuitable due to not being

available for the Netherlands, not having an open

API, or other reasons. The final choice was the Vera

Plus. In the search for suitable sensors, the

compatibility with this controller, and the in Section

3 mentioned requirements, were taken into account.

As a result, the Philiotech PST02-1A sensor was

chosen, based on its compatibility and suitability;

small, and combining both motion and door sensor.

In our test environment (see Figure 5), the

controller was placed in a location where all sensors

were still in range. The sensors were placed in each

room of interest, if possible at 1.60 m height, and

focused on the area were movement was most likely.

The sensors all monitored motion and/or

opening/closing of doors (see Section 7), and were

placed inside the appointed room for a more reliable

outcome (e.g., the toilet sensor measures motion

inside the toilet room).

Figure 5: Sensor System; showing the sensor locations.

7 DECISION SUPPORT SYSTEM

The DSS of the Server interprets the sensor data

collected from the Sensor System. Following the

various abstraction levels of the monitoring

requirements, the DSS subdivides information into

four levels: (1) raw sensor data; (2) location/action;

(3) events/alerts/statistics; and (4) trends. Information

Sensors

1. Front door

2. Toilet

3. Kitchen

4. Cutlery drawer

5. Kitchen cabinet

6. Fridge

7. Bedroom

8. Living room

9. Back door

Kitchen

Living room

Toilet

Hallway

Bedroom (upstairs)

A Remote Home Monitoring System to Support Informal Caregivers of People with Dementia

from each level is used to generate information for the

subsequent level.

At the location/action level, location refers to a

room in the house. Location cannot be tracked outside

the house, but the system can reliably detect when a

person leaves or enters the house. Actions defined at

this level are particularly cooking and sleeping.

Locations and actions are directly determined

from the door/motion sensors. We decided that the

DSS should consider the motion sensing as primary

data, as it reliably detects the presence of a person

(i.e., none of the errors were related to the motion

sensor). The door sensing is used as supplemental

data. A door opening and closing could mean a person

is either entering or leaving a room, and sometimes

doors remain open. A simple algorithm where a door

opening is assumed to mean a person leaves the room,

unless the room’s location sensor is triggered, is used.

Cooking can be determined by the fridge and on

operating either the cutlery drawer or kitchen cabinet.

We tested the location/action algorithm using the

annotated 5-day in-home test data. This test consisted

of the following annotations: leaving the house (via

front or back door); entering the house (via front or

back door); bedroom (entering or leaving); sleeping

(going to bed or going out of bed); cooking (start or

end); and toilet (entering or leaving).

For each location, we determined: minutes seen

(how many minutes being in the location were

correctly detected); minutes unseen (how many

minutes being in the location were not detected),

times seen (how many times being in the location

were correctly detected); times unseen (how many

times being in the location were never detected for the

duration the resident was there); and times false (how

many times being in the location were falsely

detected). The results are shown in Table 4.

Overall accuracy is quite high, 92%. Standing out

is the relatively high unseen minutes of the toilet. Our

findings are that the motion sensor picks up motion

slowly with respect to the typical duration of the toilet

visits (about 1-2 minutes). The number of visits is

however the most relevant for toilet use, which scores

well (only 12% total error rate). The same applies for

cooking, where times are far more relevant than

minutes (which is why minutes for cooking were not

calculated). The opposite applies to Bedroom/Sleep-

ing, Indoors (other), and Outdoors, where number of

times seen are not relevant for the alerts we generate.

For these, the number of false positives seem quite

high. This can be explained by the system registering

opening a door as leaving that room. When residents

do not actually leave the room, it takes only about 30

seconds before the motion sensor corrects that, but it

does clock up the number of false positives. This is a

trade-off between measuring minutes (necessary for

Bedroom/Sleeping, Indoors (other), and Outdoors)

and times (necessary for Toilet and Cooking).

Table 4: DSS location detection results;

Minutes seen is defined as the number of correctly detected

minutes in the location. Minutes unseen is defined as the

number of undetected minutes in the location. Times seen

is defined as the number of correctly detected times in the

location. Times unseen is defined as the number of

undetected times in the location. And times false is defined

as the number of falsely detected times in the location.

TOTAL is the sum of all above described parameters. The

corresponding percentage is calculated as follows: minutes

seen and minutes unseen are the result of dividing each by

the sum of minutes seen and minutes unseen; times seen,

times unseen, and times false are the result of dividing each

by the sum of times seen, times unseen, and times false.

Location/

action

Minutes

seen

Minutes

unseen

Times

seen

Times

unseen

Times

false

Toilet

Bedroom/

Sleeping

2637/

2271

354

Indoors

(other)

3123

140

118

Outdoors

2021

149

Cooking

TOTAL

7793

(0.92)

669

(0.08)

235

(0.69)

(0.03)

(0.28)

For the events/alerts/statistics level, the DSS

relies on the location to generate events. Most events

could be specified with the help of two types of

triggers: a time interval trigger (the resident visits a

specific location in a specific time interval); and a

timeout trigger (the resident remains in, or never

reaches, a specific location for a given duration). For

example, the resident leaving the house at night is

specified as a time trigger between 11.00 pm and 6.00

am, and the resident staying away for a long period is

specified as a timeout trigger on staying outdoors for

5 hours. In addition, the system must reliably model

the sleep/wake cycle in order to generate statistics

like sleep duration. The sleep/wake cycle progresses

when the resident goes to sleep or gets out of bed for

at least 45 minutes on particular times of the day.

Finally, the trends level uses the non-parametric

Mann-Kendall test (Kendall 1975), which enables

any trends to be detected in time-series data. The

Mann-Kendall test detects statistical significance and

direction of a trend. The test is conducted periodically

on statistics data, so the alpha value was lowered each

time using Bonferroni correction (Dunn 1961). We

did not have data to test the effectiveness of the

HEALTHINF 2018 - 11th International Conference on Health Informatics

100

algorithm, but we created simulated data to assess it.

We generated noisy data with a linear trend, with

daily values over a year, and we looked at the point at

which the algorithm could detect the trend in spite of

the noise. We used either normal or Poisson

distribution, depending on the statistics. For example,

wake-up time is normally distributed, and number of

toilet visits per day is a Poisson distribution. We

found that the algorithm could detect small trends at

low alpha levels even in cases where the trend was

not visually apparent. For example, a 30-minute shift

in wake-up time over a year with a standard deviation

of 1 hour was detected with α < 0.001. As such, we

consider the algorithm a viable proof of concept.

8 DISCUSSION & CONCLUSION

In our effort to improve HELMA with objective

monitoring, we conceptualized, developed, and

evaluated the Health Monitoring System (HMS). The

majority of identified requirements have been

implemented, and the HMS scored high in terms of

usability and quality of the Smartphone Application.

The Sensor System showed no significant flaws

during our tests, and the DSS is considered a viable

proof of concept.

The user-centered design resulted in a set of

requirements with emphasis on the provision of

relevant information in terms of the health and

independent living of the person with dementia by

monitoring their activities and behavioral changes,

and on the communication between informal

caregivers. In this way, the concept of the HMS

distinguishes itself from other systems (e.g., systems

found in systematic reviews (Ienca et al. 2017; Liu et

al. 2016; Carswell et al. 2009)) by focusing on

informal caregivers of people with dementia as

primary end-user. Moreover, the HMS aims to be a

versatile, unobtrusive, and privacy-friendly system.

The current HMS implements most of the

identified requirements. The requirements that have

not yet been addressed in this work are: outdoor

location; restlessness at night; wandering; drinking;

medication intake; personal hygiene; body weight;

falling; appliances; fire alarm; and (unwanted)

visitors. These requirements should be incrementally

added to the system to enrich the information

provision on health and independent living, and to

add the element of safety (indoors and outdoors).

Priority should be given to: outdoor location (e.g., by

using a GPS wristband); medication intake (e.g., by

using a medicine dispenser); falling (e.g., by using a

wristband or unobtrusive fall detection sensor);

appliances (e.g., by using a sensor on the stove); fire

alarm (e.g., by using a smoke detector); and

(unwanted) visitors (e.g., by registering the presence

of the visitor’s smartphone). The discussed

recommendations from participants should be

incrementally added to the Smartphone Application

to improve its functionalities and user-friendliness.

This provides particularly more overview on who

takes care of which task or alarming situation.

The Smartphone Application shows with a

median (range) SUS score of 87.50 (15.00) and

uMARS score of 4.27 (0.73) excellent usability and

good quality, suggesting that the application is most

likely well implementable in practice. The high

outcome is most likely due to our chosen design

process; involving potential end-users throughout the

design. The current drawback is the lack of testing in

a real-life setting, which could have determined

whether the provided information is sufficient, the

Calendar and Chat come to their full right, and

functionalities such as blocking calls are being used.

The Sensor System seems to work well, and the

DSS shows a high accuracy score of 92%. This is

most likely due to using relatively simple sensors and

algorithms. On the downside, the technical evaluation

was short and in a controlled environment. Future

evaluations should pursue carrying out longer (real-

life) trials, allowing to find unnoticed problems, and

to test the Sensor System on properties such as

durability and battery life. In addition, the DSS

should be brought to a higher level, using for example

machine learning to map behavior of the resident to

better identify activities and behavior, making the

system more reliable and thus more user-friendly.

The next step for the HMS would be to evaluate

the system in a real-life setting, and to search for

possibilities to merge the HMS with HELMA. We do

expect that, over time, much of the HELMA data can

be replaced with more accurate objective

measurements, making the HELMA questionnaires

more time-efficient and more focused on purely

subjective information.

In conclusion: the HMS seems to be a promising

proof of concept in providing effective support to

informal caregivers of people with dementia. The

system gives key information related to the health and

independent living of the person with dementia, and

aims therewith to promote the quality of life of the

informal caregivers by offering peace of mind and

reducing the burden of care.

A Remote Home Monitoring System to Support Informal Caregivers of People with Dementia

101

ACKNOWLEDGEMENTS

The authors would like to thank all participants,

TriviumMeulenbeltZorg, and Liberein, for their

valuable time and support. The HMS is developed by

Roessingh Research and Development, Telemedicine

Group. This work was partly funded by the H2020

program (PHC-20-2014) within the IN LIFE project

(grant no. 643442).

REFERENCES

Alzheimer’s Society, 2017. How assistive technology can

benefit people with dementia and their carers. Available

at: https://www.alzheimers.org.uk/info/20091/what_

we_think/85/assistive_technology/2.

Bank, A. et al., 2006. The value of telephone support groups

among ethnically diverse caregivers of persons with

dementia. The Gerontologist, 46(1), pp.134–138.

Benyon, D. & Macaulay, C., 2002. Scenarios and the HCI-

SE design problem. Interacting with Computers, 14(4),

pp.397–405.

Brooke, J., 1996. SUS - A quick and dirty usability scale.

In P. Jordan et al., eds. Usability Evaluation in Industry.

Taylor & Francis, pp. 189–194.

Carswell, W. et al., 2009. A review of the role of assistive

technology for people with dementia in the hours of

darkness. Technology and Health Care, 17(4), pp.281–

304.

Dunn, O., 1961. Multiple Comparisons Among Means.

Journal of the American Statistical Association,

56(293), pp.52–64.

Glascock, A. & Kutzik, D., 2000. Behavioral Telemedicine:

A New Approach to the Continuous Nonintrusive

Monitoring of Activities of Daily Living. Telemedicine

Journal, 6(1), pp.33–44.

Huis in ‘t Veld, M. et al., 2010. A scenario guideline for

designing new teletreatments: a multidisciplinary

approach. Journal of Telemedicine and Telecare, 16,

pp.302–307.

Ienca, M. et al., 2017. Intelligent Assistive Technology for

Alzheimer’s Disease and Other Dementias: A

Systematic Review. Journal of Alzheimer’s Disease,

56(4), pp.1301–1340.

Kendall, M., 1975. Rank Correlation Methods (4th ed.),

Griffin.

Kirsi, T., Hervonen, A. & Jylhä, M., 2004. Always one step

behind: husbands’ narratives about taking care of their

demented wives. Health (London), 8(2), pp.159–181.

Available at: http://www.ncbi.nlm.nih.gov/pubmed/

15068635.

Liu, L. et al., 2016. Smart homes and home health

monitoring technologies for older adults: A systematic

review. International Journal of Medical Informatics,

91(1), pp.44–59.

Ministry of Health Welfare and Sport et al., 2009. Guideline

for Integrated Dementia Care [excerpt],

Miranda-Castillo, C., Woods, B. & Orrell, M., 2013. The

needs of people with dementia living at home from user,

caregiver and professional perspectives: a cross-

sectional survey. BMC Health Services Research,

13(1), p.43.

Nielsen, J. & Landauer, T., 1993. A mathematical model of

the finding of usability problems. In Proceedings of the

INTERACT ’93 and CHI ’93 Conference on Human

Factors in Computing Systems. pp. 206–213.

Nijhof, N. et al., 2013. An evaluation of preventive sensor

technology for dementia care. Journal of Telemedicine

and Telecare, 19(2), pp.95–100.

Nolan, M., Ingram, P. & Watson, R., 2002. Working with

Family Carers of People with Dementia. SAGE, 1(1),

pp.75–93.

Peeters, J., Werkman, W. & Francke, A., 2014. Kwaliteit

van dementiezorg door de ogen van mantelzorgers.

Dementiemonitor Mantelzorg 2013: deelrapportage 1,

Peeters, J., Werkman, W. & Francke, A., 2014.

Mantelzorgers over ondersteuning bij dementie door

het sociale netwerk en de gemeente. Dementiemonitor

Mantelzorg 2013: deelrapportage 2,

Peeters, J.M. et al., 2010. Informal caregivers of persons

with dementia, their use of and needs for specific

professional support: a survey of the National Dementia

Programme. BMC Nursing, 9(1), p.9.

Peetoom, K. et al., 2015. Literature review on monitoring

technologies and their outcomes in independently

living elderly people. Disability and Rehabilitation:

Assistive Technology, 10(4), pp.271–294.

Pol, M. et al., 2016. Older People’s Perspectives Regarding

the Use of Sensor Monitoring in Their Home.

Gerontologist, 56(3), pp.485–493.

Pollitt, P., Andersont, I. & O’Connor, D., 1991. For Better

or for Worse: The Experience of Caring for an Elderly

Dementing Spouse. Ageing and Society, 11(4), pp.443–

469.

Stoyanov, S. et al., 2016. Development and Validation of

the User Version of the Mobile Application Rating

Scale (uMARS). JMIR mHealth and uHealth, 4(2),

p.e72.

Urwyler, P. et al., 2017. Cognitive impairment categorized

in community-dwelling older adults with and without

dementia using in-home sensors that recognise

activities of daily living. Scientific Reports, 7.

Available at: http://www.nature.com/articles/srep42084.

VUmc Alzheimercentrum et al., 2012. Deltaplan Dementie,

Available at: http://www.rijksoverheid.nl/documenten-

en-publicaties/rapporten/2013/04/04/rapport-deltaplan-

dementie.html.

Yang, C. & Hsu, Y., 2012. Remote monitoring and

assessment of daily activities in the home environment.

Journal of Clinical Gerontology and Geriatrics, 3(3),

pp.97–104.

Zwaanswijk, M. et al., 2013. Informal Caregivers of People

with Dementia: Problems, Needs and Support in the

Initial Stage and in Subsequent Stages of Dementia: A

Questionnaire Survey. The Open Nursing Journal, 7,

pp.6–13.

HEALTHINF 2018 - 11th International Conference on Health Informatics

102