Council of Coaches

A Novel Holistic Behavior Change Coaching Approach

Harm op den Akker

1,2

, Rieks op den Akker

, Tessa Beinema

1,2

, Oresti Banos

, Dirk Heylen

orn Bedsted

, Alison Pease

, Catherine Pelachaud

, Vicente Traver Salcedo

, Sofoklis Kyriazakos

and Hermie Hermens

1,2

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

University of Twente, Department of Biomedical Signals and Systems, Enschede, The Netherlands

University of Twente, Department of Human Media Interaction, Enschede, The Netherlands

Danish Board of Technology Foundation, Copenhagen, Denmark

University of Dundee, Dundee, United Kingdom

CNRS - ISIR, Sorbonne University, Paris, France

ITACA, Universitat Polit

ecnica de Val

encia, Val

encia, Spain

BTECH, Aarhus University, Herning, Denmark

a.pease@dundee.ac.uk, catherine.pelachaud@isir.upmc.fr, vtraver@itaca.upv.es, sofoklis@btech.au.dk

Keywords:

Virtual Coaching, Behavior Change, Embodied Agents.

Abstract:

A modern way of life needs a modern way of coaching. Despite the proliferation of ICT solutions for per-

sonalized health care, there is still no easy way to provide older adults with integrated coaching services. In

this paper we introduce the concept of Council of Coaches — a radically new virtual coaching concept based

on multiple autonomous, embodied virtual coaches, which form together a personal council that fulﬁlls the

needs of older adults in an integrated way. In this concept, coaching takes the form of an open dialog in which

clients co-construct together with a selected number of coaches their own plans to go for a healthier lifestyle.

Virtual coaches are presented to users by means of embodied conversational social characters. We discuss

technical and social challenges on the path towards realizing the Council of Coaches concept, a radically new

view of health coaching that involves the state of the art in human-computer interaction, natural dialogue, and

argumentation technology.

1 INTRODUCTION

Due to advancements in treatment and care, the aver-

age human life expectancy has increased signiﬁcantly

over the past decades. Unfortunately the added years

we gain at the end of our lives are not always lived

in good health. Instead, more and more older adults

are living under the effects of various chronic condi-

tions. Such conditions cannot be cured, but it has been

shown that a healthy lifestyle substantially contributes

to their prevention, and reduces their impact on qual-

ity of life. Adopting a healthy lifestyle means chang-

ing ingrained patterns of behavior – something that is

notoriously difﬁcult to achieve without the right sup-

port. Tools that provide personalized coaching are

rapidly emerging, and in some domains (e.g. phys-

ical activity) have penetrated the market. Key limi-

tations are that such tools focus on a single domain,

are targeted at a young and active audience, and lack

the ability to remain engaging over longer periods of

time.

In order to alleviate those issues of existing coach-

ing and support tools, we propose a radically new

concept of virtual coaching, called the “Council of

Coaches” concept. The council consists of a num-

ber of virtual characters — each specialized in their

own domain – that interact with each other and with

the user to inform, motivate and discuss about health

and well-being related issues (e.g. physical, cogni-

tive, mental and social well-being). The individual

coaches will listen to the user, ask questions, inform,

discuss between themselves, jointly set personal goals

and inspire the user to take control of his health. Any

op den Akker, H., op den Akker, R., Beinema, T., Banos, O., Heylen, D., Bedsted, B., Pease, A., Pelachaud, C., Traver Salcedo, V., Kyriazakos, S. and Hermens, H.

Council of Coaches.

DOI: 10.5220/0006787702190226

In Proceedings of the 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2018), pages 219-226

ISBN: 978-989-758-299-8

219

combination of specialized council members collab-

oratively cover a wide spectrum of lifestyle interven-

tions, with a substantial likelihood of positive impact

on outcomes related to chronic diseases.

We envision the Council of Coaches concept to

make its initial impact in the health and wellbeing

domain, especially in the multi-disciplinary case of

active and healthy aging. As such, the concept will

be demonstrated in the area of older adults suffer-

ing from age-related impairments. Additionally, we

aim to demonstrate to support the self-management

in two major chronic disorders with high volume and

high impact: Diabetes type 2 and Chronic Pain – of-

ten affecting the person in multiple health domains.

The council of virtual coaches will initiate engag-

ing discussions on healthy behaviors, forwarding their

points of view from the different domains and pro-

viding highly personalized responses on the input of

the person, using beyond state-of-the-art technology

for multi-party dialogue systems, integrated with real-

time interactive 3D embodied virtual agents. With

the Council of Coaches system, even a passive user

can be educated and convinced to take a behavior

change course-of-action by passively observing and

watching the automated discussions between virtual

coaches about the user’s behavior unfold.

Our position is that the concept of a Council of

Coaches will signiﬁcantly impact the effect of health

coaching because it (a) increases engagement, (b) al-

lows for additional coaching strategies, (c) allows to

inﬂuence an unmotivated user.

In this paper we discuss our position — a council of

autonomous virtual agents providing numerous routes

to improving collaborative support settings. We moti-

vate and further elaborate on the main concept in Sec-

tion 2. Then, in Section 3 we discuss related work,

relevant background and the inspiration that led to the

Council of Coaches Concept. In Section 4 we pro-

vide a detailed example that walks the reader through

an envisioned session with the Council of Coaches

demonstrator, explaining the concept in a story-based

scenario. In Section 5, we discuss the technical and

functional challenges we see in reaching an infras-

tructure of multiple virtual coaches cooperating with

their users in a personalized and holistic health and

wellbeing application. Finally, in Section 6 we sum-

marize our position and conclude the paper with a

view towards its realization.

2 THE COUNCIL OF COACHES

CONCEPT

Our new concept of virtual coaching involves multi-

ple autonomous agents that are able to educate and

motivate the user in interactive group discussions.

Different expert coaches will be developed in order

to demonstrate the concept, with expertise on var-

ious domains including physical activity, cognitive-

and mental health, social skills and participation, as

well as condition speciﬁc expertise for diabetes and

chronic pain.

Individual one-to-one, human-to-agent interactions

have been widely explored in research and applica-

tions in various domains. Virtual embodied conver-

sational characters for personalized coaching in the

health domain are investigated for some time already.

They focus on speciﬁc chronic conditions such as di-

abetes (Klaassen et al., 2016), or chronic obstructive

pulmonary disease (COPD), targeting better adher-

ence to medical treatment or physical activity (op den

Akker et al., 2015) or on mental diseases, drug addic-

tion (Lisetti et al., 2013), and Post-Traumatic Stress

Disorder (PTSD) (Tielman et al., 2015).

Previous studies emphasize the importance of so-

cial relationship building and companionship with the

virtual agent (Bickmore et al., 2010; Turunen et al.,

2011). Arguably the most important prerequisite for

the success of such agent-based interaction systems

is that users need to remain engaged: “if a user stops

interacting with the agent, then it cannot have any fur-

ther impact” (Battagolino and Bickmore, 2015). Be-

ing informed, and reaching motivation to change be-

havior is an internal process that needs to be boot-

strapped by engaging the user with the behavioral do-

main. By introducing the council setting of coach-

ing, we allow ourselves to keep the discussion, and

the user’s internal process of change alive, by mov-

ing the dialogue forward when the user’s initiative of

interacting with the system is low.

For example, with a traditional, single-agent coach,

the coach might ask a user (e.g. recently diagnosed

with COPD) about his intention to stop smoking,

with the underlying goal of implanting the idea that

this would be a good idea, and providing arguments

for it. However, without the user’s cooperation, a

discussion about the user’s intentions, doubts, or

difﬁculties will be impossible to have. Consider the

example below:

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

220

Coach: Have you considered trying to quit

smoking, Bob?

Bob: None of your business.

End of story. Even if a user is willing to go along

with the virtual coach, the technical limitations of

machines understanding free-form natural language

interaction severely complicates the ability of having

a meaningful, deep discussion about such topics.

Consider the following example:

Coach: Have you considered trying to quit

smoking, Bob?

Bob: Well eh, I guess the moment I walked

out of my doctor’s door, I was well it

was quite a shock I tell you.

I heard about COPD before but it’s the

type of thing you hear about other peo-

ple you know? I put my hands in my

pocket and felt the cigarette box, but

then yes I thought to myself, better not

this time. Since then I have been smok-

ing again on and off...

Coach: I did not understand your response.

Again, end of story. Now, having a group dis-

cussion where the user participates in a session

where multiple virtual agents are present allows us to

generate an interactive performance:

Ellis: Have you considered trying to quit

smoking, Bob?

Ben: Quit smoking, quit drinking, quit doing

all the fun things in life! Why don’t you

quit whining?

Florence: What a stupid thing to say, Ben! Next

thing you’re going to tell us all about

your grandfather who smoked all of his

live and lived to be a 100 years old!

Ben: Well, for your information Flo-

rence. . . My grandfather was a heavy

smoker and he did turn out to be 94

years old, so there.

Richard: Florence, Ben, cut it out, this isn’t about

you or your grandfather, we are talking

about Bob, perhaps he has anything to

say here?

Bob: (Say nothing.)

Ellis: Well, you know that smoking does not

guarantee an early death. But Bob here

recently got diagnosed with COPD, a

condition that is almost always caused

by long-term smoking.

Ellis: Although it may not be easy, it is very

important to try and quit as a very ﬁrst

step in dealing with this.

Ben: Sure, I guess, you are the expert.

Richard: And it has a positive effect on your

taste! Think about all the delicious

things we could cook together and how

excellent all the ﬂavors will work out.

Bob: I am trying but ﬁnd it difﬁcult.

What this example portrays, is that even without

any active participation of the user (Bob chooses to

say nothing), the discussion can continue and provide

some arguments and background on the topic. Then,

when the user is engaged, a predeﬁned selectable ut-

terance can indicate the stance of the user in the cur-

rent ongoing dialogue (Bob: “I am trying but ﬁnd it

difﬁcult”), allowing our COPD Coach, Ellis, to con-

tinue the dialogue with e.g. advice on how to ap-

proach the smoking cessation issue. This way, it is

possible to engage the user in a discussion about a

personal issue without bearing on real-time natural

language understanding. Furthermore, the example

illustrates some of the other underlying aims of the

project:

Immersive Dialogue — Creating a captivating di-

alogue, and unfolding stories between coaches

(Ben and Florence don’t seem to like each other

now, but who knows how romance might ﬁnd a

way?). Such stories unfolding over multiple in-

teractive session serve to further engage the user

with his personal council.

State of Art Interactive Multi-Party Conversation

— Advancing the state of the art in multi-party

dialogue systems by creating dynamic, personal-

ized dialogues that can be interrupted and driven

by user responses.

Multi-Domain Behavior Change Approach —

Approaching behavior change in a particular

domain (e.g. smoking cessation) in a holistic

way by involving in this case diet (stop smoking

positively affects taste/food experience), or

physical activity (stop smoking increases your

capacity for doing sports).

3 RELATED WORK AND

INSPIRATION

The overall concept of the Council of Coaches is in-

spired by the 1996 video game Civilization II, de-

veloped and published by MicroProse (Civilization 2,

1996), in which an in game “High Council” of ad-

visors would help the player make decisions about

strategic moves (see Figure 1). To the best of our

knowledge, no serious applications have ever been de-

veloped or researched that employ the use of a group

of advisors to educate, motivate, and engage users in

the health counseling, or in fact, any other domain.

Some examples of related research can be found

that explore applications of multiple virtual characters

Council of Coaches

221

Figure 1: Council of Advisors in Civilization II (Micro-

Prose, 1996).

in various domains. We discuss relevant examples be-

low.

Early work by Andr

e and Rist describes a sys-

tem in which multiple virtual characters are used to

present information to a user through “presentation

teams” (Andr

e and Rist, 2001) in which a virtual

car salesman and two virtual buyers display a perfor-

mance in which various aspect of the topic (car sales)

are being discussed in front of a passive viewer (the

user). The prototype described is 16 years old, and

although never further developed beyond the proto-

type, may serve as inspiration in the realization of the

Council of Coaches concept.

Traum and Rickel describe a system in which

the end-user interacts with multiple embodied vir-

tual agents in a military training scenario (Traum and

Rickel, 2002). Although many of the works techno-

logical developments done in 2002 have since been

improved, the work serves as one of the few known

examples that are comparable to the Council concept.

The authors employed a “Hollywood writer” to guide

the development by creating a storyline and script for

the training scenario. Inspired by the work of Traum,

in the Council of Coaches system we aim to integrate

creative stories that unfold between the various vir-

tual coaches throughout the users interaction in order

to enrich the individual coaches personality and in-

crease engagement with the system overall.

There is some evidence that the use of multiple

character agents can be advantageous for learners in

virtual teaching systems. In (Baylor and Ebbers,

2003) an e-learning tool was developed in which the

virtual support agent was divided into two separate

agents representing the role of teacher and that of mo-

tivator respectively. By introducing this split, an in-

crease in performance was obtained, perhaps because

the student was better able to ‘compartmentalize’ the

information provided by the two distinct coaches. In

this example the two ‘coaches’ strictly operate in a

collaborative way. In the PORTIA system — a dialog

system that allows users to engage in a natural dialog

about the relation between food and health — the is-

sue of discussion and arguments is in the center scope

(de Carolis and Mazzotta, 2017). The system dis-

tinguishes affective and rational arguments and uses

them in a persuasive dialog with the user based on a

probabilistic knowledge base and user model. Argu-

ments are presented by an Embodied Conversational

Character.

The Council of Coaches concept leans heavily on

existing knowledge and background in the ﬁeld of

coaching, personalized human-computer interaction,

embodied agents, (multi-party) dialogue and argu-

mentation theory, and behavior modeling and anal-

ysis. In the realization of this concept, we build on

the myriad of knowledge obtained and learn from the

tools and applications built and described in research.

However, a thorough discussion of the work in these

areas is beyond the scope of this position paper.

4 SCENARIO: THE STORY OF

JENNY

In order to get a tangible feeling for what the council

of coaches is all about, below we outline a detailed

scenario, that introduces our persona Jenny, a poten-

tial primary user of the system.

Jenny is 64 years old and entering the last few

months of her working life as a Sales & Marketing

Manager of a major hotel chain. Due to her life-

long desk job, she is suffering from chronic low back

pain. Jenny has just returned to her desk from a meet-

ing with the HR department to discuss her last three

months of work before retirement. Although she’s al-

ways been excited about the prospect of having more

time to work in the garden and having more time to

spend with her grand daughter, she is beginning to

worry a little about the upcoming major change in her

life. She didn’t always enjoy her working days but

she liked biking to the city, the coffee breaks with

Matthew and Abby and of course the contact with

the customers. With her ﬁnal workday rapidly ap-

proaching, Jenny is starting to worry about missing

the social contacts, getting out of the house, and get-

ting the exercise needed to get her back pain under

control. Looking through the pile of paperwork she

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

222

received from the HR manager, she ﬁnds a brochure

that catches her eye: “Worried about a major upcom-

ing life event? Go to www.council-of-coaches.eu!”.

Later that evening at home, Jenny visits the website

and reads up on the Council of Coaches application.

The website says the ﬁrst month of usage is free, so

she decides to give it a try and creates an account.

She ﬁlls in some of her user details and launches the

application. When opening the application, Jenny

is greeted by an old and wise looking man who

introduces himself as Professor Yao:

Yao: Greetings, and welcome to the coun-

cil of coaches it seems that you are in

need of some counseling, hmm? Well,

you’ve come to the right place ... I am

the head of the council of coaches, and

I have a few questions for you, if you

have some time now?

This is not what Jenny had expected, and she is

looking at the screen a little bit puzzled.

Yao: Hmm? Well, do you have some time?

“All right, I have some time now”, Jenny says,

and the grey old ﬁgure on the screen continues.

Yao asks a number of questions about what’s trou-

bling Jenny, and ﬁnally stops to think for a little while.

Yao: Based on what you say I suggest follow-

ing. Do not worry Jenny we have a team

of experts here to help you.

I recommend you discuss your worries

about your physical activity with Sgt

Brady, he is a retired army drill instruc-

tor, but I’m sure you will ﬁnd him to be

reasonably pleasant enough.

Next I recommend you talk about your

social worries with Sonja, or “Sociable

Sonja” as she likes to call herself. She

will have some good advice on how to

make sure you’re not isolating yourself.

And ﬁnally, I recommend you talk to

Pete. Pete was recently retired from

the council of coaches, but he is hav-

ing troubles letting go of his duties. I’m

sure he understands well the issues that

you’re facing concerning your retire-

ment.

Finally, coach Tracy is an expert on

chronic pain, I’m sure you’ll ﬁnd her to

be very helpful.

Yao wishes Jenny the best of luck and tells her that

she can call on him whenever she needs him. Next,

the scene changes to the council chamber, where the

three coaches that Yao introduced are waiting for

Jenny.

Brady: So you are afraid of becoming unﬁt for

duty?! Start exercising, walking, cy-

cling! In fact, give me 100 push ups

right now!

Sonja: Whoa there, Brady, take a deep breath.

We’ve not even been introduced! That

was Sgt Brady and you have to excuse

his enthusiasm, he hasn’t had the oppor-

tunity to coach anyone for a long time.

Besides, we don’t even know how ac-

tive Jenny is.

While everyone on the council is looking at Jenny,

a short physical activity stage of change questionnaire

appears on the screen. Jenny indicates that she hasn’t

been very active but that she intends to be.

Brady: That’s good to know, but I need more

information to work with. If you’re

planning to become more physically ac-

tive, I need to know how active you are

now. Do you have an activity tracking

device, Jenny?

Jenny indicates that she hasn’t got one yet, but that

she may look for one in town on Saturday. The dis-

cussion between the council members continues for

some time as they are discussing what seems to be

the right approach for Jenny. Tracy makes sure the

other coaches take into account Jenny’s chronic pain.

Jenny listens and intervenes every now and then, until

she decides that she has to go. She says goodbye to

the council, and the council members wish her a nice

day and good luck. Curious as to what Sgt Brady will

have to say, she decides to bike to the local electronics

store to buy an activity tracker, and plans to talk to the

council again later that evening.

Three months later... Jenny is now ofﬁcially re-

tired. In the last three months she has used the Coun-

cil of Coaches application on her laptop. Pete, the

retirement coach, has especially given her some use-

ful advice on ﬁnding some outdoor hobbies and she

joined a weekend walking club in the area. She has

been using her physical activity tracker to keep track

of her activity and she set a personal step goal for her-

self together with Sgt Brady and Tracy. In her last

meeting she was advised to download the mobile app

for her smartphone so that the coaches could reach

her when she was on the move, and she decided to

install it. Having the app on her smartphone allows

Jenny to contact her personal coaches whenever and

wherever she decides. Jenny’s phonebook includes a

new entry for Sgt Brady, Sonja, Pete, and Tracy, and

Council of Coaches

223

talking to them is as easy as making a phone call. She

can also text them whenever she is in a more public

environment. In these last three months, Jenny grew

a fond affection for the coaches. She feels that each

coach cares about her health, and makes her sympa-

thize with them. Even Sgt Brady can have his good

days.

5 CHALLENGES

In order to realize the Council of Coaches concept,

a number of technical challenges need to be over-

come. In Figure 2 we sketch a high-level concep-

tual overview of an envisioned implementation of the

Council of Coaches concept, highlighted further in

the paragraphs below.

Figure 2: Council of Coaches — Conceptual Overview.

5.1 Coaching Strategies and Knowledge

Design

Moving beyond existing literature on coaching strate-

gies, in the Council of Coaches concept, the presence

of multiple coaches allows for approaches in which

coaches either apply separate coaching strategies at

the same time or join together in applying one strat-

egy. Examples of such a multi-coach strategy could

be ‘good coach, bad coach’, or a ‘set up’ to elicit in-

formation from the user (as in one of the previous ex-

amples where Sonja uses Sgt Brady’s ‘Give me 10

push ups!’ statement to ask Jenny about her activity

stage). Deﬁning these joined strategies as a general

scheme so that they can be adopted by coaches when

desired could be a possible implementation. For ex-

ample, a simple scheme to elicit information could

be: ‘Coach 1: Makes a controversial statement.’ and

‘Coach 2: Moderates the statement, and asks for in-

formation.’.

Having multiple coaches in the council that each

have their own expertise also allows for the introduc-

tion of new coaches when their expertise is required.

Plugging in a new coach will introduce new topics

for conversation and this will require that the other

coaches also share some knowledge on the subject.

Representing this knowledge and the coaches in such

as way that they can be added as modules in the sys-

tem could facilitate this process.

5.2 Holistic Sensing

Having a clear understanding of the user behavior

is essential to elicit effective and relevant coaching

strategies. Despite the fact that signiﬁcant progress

has been made in the sensing and quantiﬁcation of

behavior in recent years, the vast majority of exist-

ing solutions are domain speciﬁc, i.e., focusing ex-

clusively on either physical, emotional, cognitive or

social aspects. More importantly, most applications

claim to infer human behavior in a broad sense when

they merely detect some activity-related parameters,

such as step counts, calories burned or ﬂoors climbed.

In order to realize the Council of Coaches con-

cept, a multi-domain sensing and behavior analysis

framework must be realized that can feed the neces-

sary knowledge bases of the individual agents. This

knowledge will provide relevant background informa-

tion for improving the user-coach interaction and sup-

porting the dialogue and argumentation process with

objective and continuous behavior-related contents.

The concept moves beyond current trends in auto-

matic behavior analysis while approaching the prob-

lem in a holistic fashion, thus modeling, inferring and

combining the distinct domains of behavior including

physical, emotional, cognitive and social aspects.

5.3 Dialogue and Argumentation Design

One of the advantages of having multiple coaches

is the ability to apply the concept of ‘dialogue as

performance’. That is, the coaches are able to per-

suade and re-engage a (passive) user by performing,

in essence, a play. This opens up new approaches to

persuasion dialogues, with research questions such as

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

224

how to model a “silent user” and whether “passive

participation” is effective in engaging users. A ma-

jor challenge lies in enabling such, essentially agent-

agent, dialogues that are able to convince the non-

participating partner. Both from the perspective of

coaching and the perspective of argumentation tech-

nology, this requires further study.

Another challenge presents itself in developing

new dialogue types, in particular facilitation. Mo-

tivation to change should be elicited from the user,

rather than being imposed by the coaches — an op-

portunity particularly made possible by the Council

of Coaches setting. This focus on facilitating deci-

sions, as opposed to direct persuasion, differs from

standard persuasion dialogue models, with coaching

offering a new domain for dialogue theorists. It will

open up new research questions such as how to have a

generally amicable dialogue (e.g. deliberation) which

has as its intention an outcome which is generally as-

sociated with an adversarial kind (e.g. persuasion).

Having dynamic strategies, in which the coaches

adjust their conversational strategies to the unfold-

ing dialogue, will also be considered. In highlight-

ing the social behavior via multiple virtual agents, the

project will open the way for both incorporating ﬁnd-

ings from, and adding to, social psychology concern-

ing group dynamics, dominant agents and the role of

a moderator (and whether there should be one).

5.4 Human-Computer Interaction

Anthropomorphic virtual agents are designed as sim-

ulated social role models in various domains and ap-

plications, among which the role of a life style coach

or therapist in the health domain. Several studies have

shown that the visual presence and appearance of such

agents have a major impact on motivation and affect

(Baylor et al., 2009). In order to be effective in per-

suading the user, a coach should use affective as well

as rational arguments and adapt its persuasive strategy

to the goals and value preferences of the user. As mul-

tiple coaches introduce additional challenges in terms

of the Coaching Strategies (Section 5.1) and the dia-

logue (Section 5.3), so too new challenges arise in the

visual representation of a group of embodied agents.

Different models enabling Embodied Conversa-

tional Agents (ECAs) to exhibit social attitudes

through their verbal and non-verbal behavior have

been proposed. For instance, in (Gillies et al., 2004),

the system Demeanor supported the design of virtual

characters within a group with different social atti-

tudes (expressed through posture and gaze). In (Rehm

and Endrass, 2009), the authors proposed a toolbox

for manipulating the decision making of agents in a

group based on different theories on social relations.

In (Lee and Marsella, 2011), the authors designed

the nonverbal behavior of Embodied Conversational

Agents depending on their conversational role and

their social relations. In (Cafaro et al., 2012), the

authors conducted a study where users evaluated the

perception of an ECA’s attitude (friendly, hostile, ex-

traversion) in the ﬁrst seconds of an encounter with

an ECA exhibiting different behaviors (smiles, gaze

and proxemics). As an emerging ﬁeld, the Council

of Coaches concept is expected to build on and con-

tribute signiﬁcantly to the area of multi-party embod-

ied interaction.

5.5 Responsible Research and

Innovation

Finally, the Council of Coaches concept introduces

radically new ways of interaction between digital sys-

tems and users. This innovative interaction will be

demonstrated in the area of health and wellbeing for

elderly users — a target group that can be considered

especially vulnerable. For these reasons it is impera-

tive that the research and innovation process follows

the principles of Responsible Research and Innova-

tion (RRI). In the realization of this concept, we need

to make sure that the new tools and coaching meth-

ods, as well as the actual R&I processes contribute

effectively to the needs of future users and society at

large, are aligned with societal values and ethical con-

siderations, and have a realistic chance of succeeding

in the competitive market of e-Health and e-coaching

tools.

6 CONCLUSIONS

Despite the proliferation of ICT solutions for person-

alized health-care, there is still no easy way to pro-

vide older adults with integrated coaching services.

Council of Coaches introduces a radically new vir-

tual coaching concept based on multiple autonomous,

embodied virtual coaches, which form together a per-

sonal council that fulﬁlls the needs of older adults in

an integrated way. Each coach has his own expertise,

personality and style of coaching; they might not al-

ways agree with each other, but they all share a single

goal: to support the user across every aspect of well-

being, including physical, social, cognitive and men-

tal support. The Council members listen to the user,

Council of Coaches

225

inform the user, help the user to set personal goals

and inspire him to take control of his health. Give the

council your thoughts, or listen and observe how the

individual coaches exchange their views on numerous

issues. Apply your new skills in daily life and if the

need arises, contact any of the coaches anytime, any-

where. . .

The ﬁelds of behavior change, health information

provision, e-learning, and human-computer interac-

tion in general will greatly beneﬁt from the Council

of Coaches approach. By developing a framework in

which multiple, completely autonomous agents can

hold interactive dialogues with a user, we open up

the possibility for numerous new and exciting innova-

tions in those research ﬁelds. We presented the overall

concept and the major technological and social chal-

lenges in realizing the Council of Coaches and will

take the necessary steps to overcome these challenges

in the next three years.

ACKNOWLEDGEMENTS

The European Commission has provided funding for

realizing the Council of Coaches concept in the same-

titled project under Grant Agreement No. 769553.

REFERENCES

Andr

e, E. and Rist, T. (2001). Presenting through per-

forming: On the use of multiple lifelike characters in

knowledge-based presentation systems. Knowledge-

Based Syst., 14(12):313.

Battagolino, C. and Bickmore, T. (2015). Increasing

the engagement of conversational agents through co-

constructed storytelling. In Intelligent Narrative Tech-

nologies and Social Believability in Games: Papers

from the AIIDE 2015 Joint Workshop.

Baylor, A. L., Arellano, D., Varona, J., Perales, F. J., Bailen-

son, J. N., et al. (2009). Promoting motivation with

virtual agents and avatars: role of visual presence and

appearance. Philosophical transactions of the Royal

Society of London. Series B, Biological sciences.

Baylor, A. L. and Ebbers, S. (2003). Evidence that multiple

agents facilitate greater learning. In Artiﬁcial Intel-

ligence in Education: Shaping the Future of Learn-

ing Through Intelligent Technologies; U. Hoppe, M.F.

Verdejo, J. Kay (Eds.), pages 377–379. IOS Press.

Bickmore, T. W., Puskar, K., Schlenk, E. A., Pfeifer, L. M.,

and Sereika, S. M. (2010). Maintaining reality: Rela-

tional agents for antipsychotic medication adherence.

Interacting with Computers, 22(4):276 – 288. Sup-

portive Interaction: Computer Interventions for Men-

tal Health.

Cafaro, A., Vilhj

almsson, H. H., Bickmore, T., Heylen, D.,

ohannsd

ottir, K. R., and Valgarosson, G. S. (2012).

First impressions: Users’ judgments of virtual agents’

personality and interpersonal attitude in ﬁrst encoun-

ters. In Lecture Notes in Computer Science (including

subseries Lecture Notes in Artiﬁcial Intelligence and

Lecture Notes in Bioinformatics).

de Carolis, B. and Mazzotta, I. (2017). A user-adaptive per-

suasive system based on a-rational theory. Interna-

tional Journal of Human-Computer Studies, 108(Sup-

plement C):70 – 88.

Gillies, M., Crabtree, I. B., and Ballin, D. (2004). Cus-

tomisation and context for expressive behaviour in the

broadband world. BT Technology Journal.

Klaassen, R., op den Akker, R., Di Bitonto, P., van der Burg,

G.-J., Bul, K., and Kato, P. (2016). Pergamon: A se-

rious gaming and digital coaching platform support-

ing patients and healthcare professionals. In Varajao,

J., Cruz-Cunha, M., Martinho, R., Rijo, R., Bjorn-

Andersen, N., Turner, R., and Alves, D., editors, In-

tern. Conf. ENTERprise Information Systems/Intern.

Conf. Project MANagement/Intern. Conf. Health and

Social Care Inform.Systems and Techn., 2016; CEN-

TERIS/ProjMAN/HCist, pages 261–269. SciKA.

Lee, J. and Marsella, S. (2011). Modeling side participants

and bystanders: The importance of being a laugh

track. In Lecture Notes in Computer Science (includ-

ing subseries Lecture Notes in Artiﬁcial Intelligence

and Lecture Notes in Bioinformatics).

Lisetti, C., Amini, R., Yasavur, U., and Rishe, N. (2013). I

can help you change! an empathic virtual agent deliv-

ers behavior change health interventions. ACM Trans.

Manage. Inf. Syst., 4(4):19:1–19:28.

op den Akker, H., Cabrita, M., op den Akker, R., Jones,

V. M., and Hermens, H. J. (2015). Tailored moti-

vational message generation: A model and practical

framework for real-time physical activity coaching.

Journal of biomedical informatics, 55:104–115.

Rehm, M. and Endrass, B. (2009). Rapid prototyping of

social group dynamics in multiagent systems. AI &

SOCIETY.

Tielman, M., van Meggelen, M., M.A., N., and Brinkman,

W. (2015). An ontology-based question system for

a virtual coach assisting in trauma recollection. In

Brinkman WP., Broekens J., Heylen D. (eds) Intelli-

gent Virtual Agents, IVA 2015. Lecture Notes in Com-

puter Science, vol 9238. Springer.

Traum, D. and Rickel, J. (2002). Embodied agents for

multi-party dialogue in immersive virtual worlds. In

Proceedings of the First International Joint Confer-

ence on Autonomous Agents and Multiagent Systems:

Part 2, AAMAS ’02, pages 766–773, New York, NY,

USA. ACM.

Turunen, M., Hakulinen, J., Stahl, O., Gamback, B.,

Hansen, P., Gancedo, M. C. R., de la Cmara, R. S.,

Smith, C., Charlton, D., and Cavazza, M. (2011).

Multimodal and mobile conversational health and ﬁt-

ness companions. Computer Speech & Language,

25(2):192 – 209.

ICT4AWE 2018 - 4th International Conference on Information and Communication Technologies for Ageing Well and e-Health

226