An Adaptive, Structural and Content Gamiﬁcation Concept for

Regulated Daily Routines

Martin Franke, Bianca Zimmer and Thomas Schlegel

Technische Universit

at Dresden, N

ohnitzer Str. 46, D-01187 Dresden, Germany

Keywords:

Assisted Living, Anomaly Detection, Context Awareness, Quality of Life, Gamiﬁcation.

Abstract:

In this work, we introduce our concept for regulating daily routines in assisted living environments by the

use of gamiﬁcation elements. Nowadays mobile devices and sensors are ubiquitous, so they are suitable

for assisting users on reaching their personal goals. One of the most pressing challenges in this regard is

the preservation of long-term user motivation. In this paper we propose a gamiﬁcation concept for a mobile

application that support the achievement of regulated daily routines. With our assistive knowledge- and model-

based approach SeMiWa, the system detects irregularities and their potential cause. These obtained insights

are used to adapt the content of the application in a gamiﬁed way. A prototypical implementation substantiates

our approach and introduces our forthcoming user study.

1 INTRODUCTION

Regular daily routines not only increase the over-

all well-being, but also affect peoples’ health and

the perceived stress level (Minors and Waterhouse,

2013). Large discrepancies in these rhythms may

result in sleep disorder up to chronic depressions

(Spork, 2011). Nowadays, daily routines can only be

manually supervised by experts in order to avoid these

negative consequences. The intention of our system is

to offer automated and preventive assistance for this

complex process. As one solution, Ambient Assisted

Living (AAL) systems have been proposed (Klein-

berger et al., 2007). These systems aim to assist users

in their domestic environment to improve overall life

quality.

One type of such systems - coming of the do-

main of elderly care - features domestic living envi-

ronments with activity recognition. These systems

monitor surrounding environments and user activi-

ties in order to ensure elderly are living safely in

their own homes while staying independent in their

life style. By utilising activity recognition algorithms

such systems are able to detect a range of activities,

e.g., walking, sleeping, and eating (Chernbumroong

et al., 2013; Fleury et al., 2010). For this purpose,

supervised machine learning algorithms are used to

build training sets. However, beside the classiﬁca-

tion of uncorrelated activities, it becomes more and

more necessary to combine these activities to daily

routines. Since a permanent interruption of routines

could lead to diseases, ranging from sleeping troubles

up to chronical depression. Studies have shown that

people, who live in regular daily rhythms, are less af-

fected by these diseases (Spork, 2011).

We developed a concept that detects deviations in

daily routines and tries to ﬁx potential provenances

with the help of gamiﬁcation.

Moreover, in the initial – and sometimes during

runtime – the application depends on manual user in-

put, we motivate the user preliminarily also with the

use of gamiﬁcation elements.

In the following sections we detect common

used game elements of mobile applications for self-

improvement in terms of their motivational effect.

Section 3 presents an overview of our model-based

approach SeMiWa that analyses sensor events as well

as performs the recognition of activities and daily rou-

tines. Section 4 outlines our gamiﬁcation concept de-

veloped for reaching personal goals and the ﬁrst pro-

totypical application. Section 5 concludes this pa-

per with a discussion and shows directions for future

work.

2 RELATED WORK

We identiﬁed approaches related to our work that can

be classiﬁed into research of chronobiology as well

233

Franke M., Zimmer B. and Schlegel T..

An Adaptive, Structural and Content Gamiﬁcation Concept for Regulated Daily Routines.

DOI: 10.5220/0005317802330240

In Proceedings of the International Conference on Biomedical Electronics and Devices (SmartMedDev-2015), pages 233-240

ISBN: 978-989-758-071-0

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

as approaches for gamiﬁcation. As we discussed pro-

foundly the state of the art in assisted living envi-

ronments (Franke et al., 2013b; Franke et al., 2014b;

Franke et al., 2014d), we omit these ﬁeld in this work.

2.1 Regulated Daily Routines

Daily routines not only depend on personal mat-

ters, but also on exogenous and endogenous factors.

In general cases, these routines evolve on natural

rhythms of one person. For instance, measures from

light exposure sensors are showing evidence in the

sleep-awake rhythm.

Also the change of temperature that affect the

overall activity of people. Changes in the atmospheric

pressure can lead to headaches in some people and

rainy conditions can reduce motivation to go outside.

In contrast, spring time can lead to an extreme raise of

activities until night time (Hildebrandt et al., 2013).

The ﬁeld of chronobiology (Hildebrandt et al.,

2013; Zulley and Knab, 2013) examines these im-

pacts to our biological clock, such as typical sleeping

cycle or times of high physical ﬁtness. A distinction is

made between circadian, ultradianen, and infradian

rhythms.

The circadian rhythms last for nearly one day with

exactly one peak and one low in this duration. Typical

examples are the sleep-waking rhythm, overall physi-

cal and mental capacities or the hormone production.

The ultradian rhythms remains for a few hours. Char-

acteristic examples are eating and drinking or the raise

and lower of blood pressure. The infradian rhythms

continues beyond the day limit. These rhythms can be

as long as the seasonal affective disorder (SAD) that

lasts for several months or the women’s menstruation

cycle with a peak once a month.

All these rhythms have in common that they di-

rectly effect humans’ behaviour. For instance, in the

SAD it is likely to eat more food or being less at-

tracted to go outside. We formalised rules out of

these facts, ﬁrstly for the activity recogniser and sec-

ondly for giving healthy advices, such as eight hours

of sleep or performing a “good” amount of sport ac-

tivities.

2.2 Gamiﬁcation

The ﬁeld of Gamiﬁcation describes the application

of game design elements in non-game contexts (De-

terding et al., 2011). The success of existing learn-

ing and self-tracking apps (e.g. the language trainer

Duolingo

, the running app Nike+

or Withings

App) conﬁrms that gamiﬁcation can be an effec-

tive tool to motivate user activity and long-term use.

Gamiﬁed apps can support users to consistently pur-

sue personal projects and goals. User motivation is

generated by various game elements used in self-

improvement apps. These elements can be deﬁned

as speciﬁc characteristics of games that can be ap-

plied in gamiﬁcation (Deterding et al., 2011; Werbach

and Hunter, 2012). In the literature, game elements

are often introduced on different levels of abstraction.

The game component points for example is a numer-

ical representation of the games progression and can

therefore be considered as a more- speciﬁc form of

the element progression (Werbach and Hunter, 2012).

Kapp et al. (Kapp, 2013) distinguish between two

general types of gamiﬁcation: structural and content

gamiﬁcation. Structural gamiﬁcation on the one hand

means that the structure around the content becomes

game-like to motivate users to go through the content.

Exemplary elements of structural gamiﬁcation can be

points, badges or leaderboards. On the other hand

content gamiﬁcation alters the content, e.g. by means

of avatars and challenges, to make it more game-like.

According to Kapp et al. (Kapp, 2013) the combina-

tion of both, structural and content gamiﬁcation, is the

most effective way to increase the users motivation.

In a literature review Hamari et al. (Hamari et al.,

2014) gathered different game elements as motiva-

tional affordances tested in empirical studies, iden-

tifying points, leaderboards and badges as the most

commonly found elements. However, the question

rises, which elements have a particularly motivating

effect on users.

There are already a few approaches to gamify as-

sisted living systems (Burmeister et al., 2013), but to

the best of our knowledge, a usage of a combination in

structural and content gamiﬁcation is not used within

such systems so far.

3 SeMiWa - SEMANTIC

MIDDLEWARE

To unify and process heterogeneous sensors in as-

sisted living environments, we use our approach

called Semantic Middleware (SeMiWa) (Franke et al.,

2013a). This model-based, event-driven middleware

acts as an intelligent discovery and routing service for

events. The middleware is parted into a semantic stor-

http://www.duolingo.com

http://nikeplus.nike.com

http://www.withings.com

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

234

Figure 1: SeMiWa Overview (Franke et al., 2014a).

age (MDRO), a semantic processing unit (Annotation

Component) and a network transparent interface for

data transmission (Perception), illustrated in Figure

refﬁg:semiwa.

3.1 Architecture Overview

The SeMiWa’s Knowledge Base provides a central se-

mantic storage for all semantic data within the system,

such as sensor models, activity patterns, routine mod-

els, publish-subscribe (PubSub) registration and life-

cycle information about all registered Nodes (Franke

et al., 2014b).

The Annotation component annotates all trans-

ferred data (i.e. register messages, sensor values and

PubSub patterns) according to our models situated in

the knowledge base. This detection, respectively clas-

siﬁcation, is done by using meta-information about

the sensor in the register messages (Tan et al., 2012).

These register messages consist of identity informa-

tion (unique sensor id), the classiﬁcation information

(sensor type, model and unit) and a timestamp for life-

cycle purposes. As we work on predeﬁned sensor

models, only one exact match is found and no ma-

chine learning approaches has to be applied for this

semantic enrichment. Furthermore, all sensor values

are used to enhance them to user’s activities (Franke

et al., 2013b; Franke et al., 2014a).

The Perception component opens a uniquely iden-

tiﬁed, bidirectional, network transparent channel for

Node registration and events. By means of the knowl-

edge, coming from the Knowledge Base, it routes

events to subscribers (PubSub) and service calls from

invoking Nodes to selected ones (RPC). RPC calls

are services on the selected sensors, such as conﬁg-

ure the location of the weather sensor or changing its

frequency.

Our main concept is to handle and describe all

possible elements, such as applications, sensors or ac-

tuators, uniformly as Nodes.

3.2 Modular Daily Routine Ontology

The base of our activity recognition approach is a for-

malised Modular Daily Routine Ontology (MDRO)

(Fig. 2) (Franke et al., 2014b). It provides a RDF-

S/OWL vocabulary for annotating data sources, such

as sensor values, factual knowledge from Chronobi-

ology, recognised activities and routines. The four

MDRO modules (sensor, activity, routine and user)

represent different facets of the activity recognition

domain. They refer to each other and to other existing

ontologies as needed. In the following sections, we

describe the essential parts of the ontology, which are

used for recognising current activities and determin-

ing daily routines.

3.2.1 Sensor Module

The sensor module provides a semantic description

for all sensors, provided by an ontology derived from

the StarFL sensor ontology (Malewski et al., 2012),

extended by a combination with QUDT

and OWL-

. Therefore, it is possible to semantically unify sen-

sor values and data types within the overall system.

Instead of broadcasting raw integer or double values,

all transmitted sensor data is annotated with seman-

tic information, such as “Temperature data in degC”

or “Humidity in %” thus providing knowledge for the

system how the raw sensor values can be interpreted.

A full list of properties can be found in (Franke

et al., 2014b). The most important feature regarding

this work is the healthyInterval. These intervals are

formalised from the ﬁeld of chronobiology. For in-

stance, we abstracted a light exposure interval, a step

count interval of more than 10000 steps or a body

mass index between 18.5 and 25. These properties

are used later for the adaptive content gamiﬁcation

(cf. Sect. 4).

http://www.qudt.org

http://www.w3.org/Submission/OWL-S/

AnAdaptive,StructuralandContentGamificationConceptforRegulatedDailyRoutines

235

3.2.2 Activity Module

The activity module holds information about recog-

nisable Activities of daily living (ADL). As initial, but

extendable set, we used the caring descriptions, also

mentioned as basic ADLs (Fleury et al., 2010) and

transferred them in a hierarchical semantic model.

Furthermore, this module stores the semantic

training set. Every classiﬁed activity is linked to one

or more sensors from the sensor module and with the

corresponding sensor values. This is necessary to jus-

tify the classiﬁed activity to the user, since we do not

use all sensors at anytime for classiﬁcation (Franke

et al., 2014b).

Regarding this work, we extend the classiﬁable

activities by the properties healthyDuration, tooLow-

Duration and tooHighDuration. These durations are

stored in generic classes, such as SportsActivity, to

sum up all derived sub classes (Walking, Hiking,

Swimming), in case no different duration is given.

Some examples of these durations are Sleep (per day)

between 6.5h & 9.5h and an optimum at 8h and Sport

(per week) between 2h & 7h and an optimum at 5h.

3.2.3 Routine Module

The routine module saves a directed chain of ADLs

with corresponding transition probabilities. Every

part of a routine is a unique instance of the activity

module, combined with the daytime. The probabili-

ties are calculated during the user’s history. For exam-

ple, four days with transitions from Sleep to Breakfast

and one with a transition to Work will result in transi-

tions of 80% and 20%.

As a common vocabulary we utilised the Per-

sonal Information Model Ontology (PIMO) (Sauer-

mann et al., 2007), as it is the quasi standard for se-

mantically describing daytime sequences. With this

ontology, we can describe duration, start and end of

one activity (hasActivity). This combination keeps a

chain through our activity module and works in con-

junction with other calendar-based applications due to

the usage of the PIMO standard.

Based on these semantically saved routines, we

can perform the deviation and cause detection, as our

ontology keeps the chain to our activity module and

moreover to the sensor module (cf. Sect. 3.5).

3.3 Activity Recognition Algorithm

The activity recognition algorithm creates an ordered

list of possible detected activities for the current sen-

sor snapshot. The algorithm ranks the possibility of

one activity according to the current time. This sort-

ing is assigned by factual knowledge; classiﬁed sen-

Figure 2: Modular Daily Routine Ontology (Franke et al.,

2014b).

Figure 3: Sensor selection (Franke et al., 2014a).

sor values; and user generated knowledge (Franke

et al., 2013b; Franke et al., 2014a; Franke et al.,

2014d). The Equation 1 illustrates its fundamentals:

P =

n := {2,3}

∗ r

+ r

(+r

)) (1)

1) Factual Activity Knowledge. The factual activ-

ity (r

) knowledge is deﬁned by a set of rules from

Chronobiology, e.g., veriﬁed statements such as “in-

fradian sleep cycles”. These rules consist of a condi-

tion, e.g., time of day, and a probability (Franke et al.,

2014b).

2) Classiﬁed Sensor Values. The classiﬁcation

probability (r

) is calculated by our classiﬁer ac-

cording the current sensor information. The ac-

tivity recognition, respectively the routine recogni-

tion, reuses approaches based on Support Vector Ma-

chines (Fleury et al., 2010; Chernbumroong et al.,

2013).

3) User-shared Knowledge. The factor r

is as-

sociated with our concept of employing the user-

generated knowledge through using collaborative ﬁl-

tering (Franke et al., 2014d). It is either the previous

activity, done at the same time or the algorithm tries

to foresee a possible rating based on similar users.

The meaning factor n is assigned with n = 2 if nei-

ther a previous activity at same time can be found, nor

a prediction can be calculated. In all other cases it is

assigned with n = 3 to keep the equivalently rating of

all factors.

Finally, the complete list of activities is ordered

based on combined probabilities P for each mapping.

This ranking is used to assign the most probable ac-

tivity to the user’s daily routine.

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

236

3.4 Sensor Selection

For our purposes, we use a selection of sensors and

devices to stay mostly comparable to nowadays activ-

ity recogniser systems (see Figure 3) (Franke et al.,

2013b; Franke et al., 2014a; Franke et al., 2014c):

Physical Sensors. We abstract physical, climatic and

surrounding sensors out of the NetAtmo

sensor

station. Each station consists of a temperature, a

humidity and a CO2 sensor. Additionally, we can

use one barometer and sound sensor in the main

room. All sensors are assigned to a speciﬁc loca-

tion that is described by spatial coordinates in the

domestic environment. This information is later

used to reduce the dataset according to the classi-

ﬁcation and cause detection (Franke et al., 2014c).

Location Sensor. With the aid of the Moves

app

and API, we built a location sensor, which records

basic activities (walking, cycling, transportation

and running) and the user’s locations (Franke

et al., 2013b).

Body Sensors. Besides the Moves app, we use two

Withings

devices for health tracking. First of

all, the Withings Pulse for measuring covered

distances, sleep cycles and pulse measures, and

second, the Withings Smart Scale for measuring

weight and body fat mass.

Weather Sensor. We wrapped the WeatherUnder-

ground

service, which provides information

about temperature, humidity, air pressure, lumi-

nance, and in general the weather on the users lo-

cation.

Calendar Sensor. As ’hint generator’ for the classi-

ﬁer, we abstracted an online calendar, which pub-

lishes calendar events predeﬁned by the user.

3.5 Deviation Detection

To acquire an active assistance for the user, the pro-

posed system detects causes of anomalies proactively,

and reactively. The reactive anomaly detection recog-

nises anomalies in one chosen feature at a particular

time interval, e.g., inﬂuencing factors of sleep dura-

tion last week and the proactive anomaly detection

recognises anomalies in the daily routines during run-

time, based on different time slots (e.g., week, month,

quarter). These anomalies were directly presented to

http://www.netatmo.org

http://moves-app.com

http://withings.com

http://wunderground.com

the user in our previous work with the help of an as-

sistive mobile application. For the relevant technical

details on deviation detection consider our previous

work (Franke et al., 2014c).

4 GAMIFICATION CONCEPT

FOR REGULATED DAILY

ROUTINES

First of all, in order to compare commonly used game

elements with regard to their motivational effect to

the user, we designed an online survey as preliminary

study. After analysing different existing gamiﬁed

apps for self-improvement, we identiﬁed frequently

recurring game elements that had to be assessed with

respect to their motivational effect on the participants.

Thereby, commonly used elements of both, structural

and content gamiﬁcation, were considered (cf. Figure

4).

By integrating screenshots of selected examples of

gamiﬁed apps for self-improvement into the survey,

we evaluate the central question: “Which game ele-

ments in this app motivate you to achieve your goal?”

At the time of writing, a total of more than 40 par-

ticipants completed the survey. The age of the partic-

ipants varies from 13 to 55+ years, 63% female and

37% male. The majority of the participants (91%) are

registered at one or more social networks and 84%

of them are using Facebook. This aspect is relevant,

since more and more gamiﬁed apps are integrating so-

cial networks for social sharing or competing against

friends.

As an introduction to each section of the question-

naire, the relevant app was presented. The main ques-

tions are related to the personal evaluation of the mo-

tivational affordances using assessment tables (from

“totally motivating”, “rather motivating”, “rather not

motivating”, “not motivating at all” or “I dont know”).

The participants answers to the questions are pre-

sented in Fig. 4. Of particular interest is that only

2% of all respondents agree that sharing achievements

through social networks (cf. Social Sharing in Fig. 4)

is motivating, although most of the participants are

registered at a social network. However, over 80% are

convinced that challenging and varied tasks are moti-

vating. Furthermore, the game components content

unlocking with a score of 79%, progress indication

and levelling up with 77% approval achieved similar

good results.

From these preliminary results it can be con-

cluded, that progress indication and levelling up are

particularly motivating game elements of structural

AnAdaptive,StructuralandContentGamificationConceptforRegulatedDailyRoutines

237

Figure 4: Long-term motivation comparison of varying

gamiﬁcation elements.

gamiﬁcation. Similarly, challenging and varied tasks

as well as content unlocking have a motivational ef-

fect for the majority of the respondents. Conse-

quently, the higher-level elements challenge and cu-

riosity (Kapp, 2013) can be identiﬁed as important

motivational affordances in the ﬁeld of content gami-

ﬁcation. It should be noted that the presented ques-

tionnaire results were averaged over various apps.

When regarded individually, it is evident that the same

game elements (e.g. leaderboards) were assessed dif-

ferently for particular apps. This indicates that the

motivating effect of game elements is also inﬂuenced

by other factors. Consequently, subsequent studies

need to consider additional aspects, e.g. demographic

user data, the visual design of game elements and

other contextual criteria. However, one usable trend is

already evident: not all game elements have the same

motivating effect.

In order to apply gamiﬁcation concepts, we are de-

veloping a mobile advisory application in the ﬁeld of

regulated daily routines. In Figure 5, a screenshot of

the prototypical mobile application “SeMiWa Watch-

board” is illustrated. With interactive gamiﬁed activ-

ities around adaptive, weekly changing subjects (e.g.

sleep and sport durations) the user can adapt, moni-

tor and evaluate his own lifestyle habits. By master-

ing a weekly challenge the user is confronted with a

balanced work-life, sleep-wake, or sport amount in a

playful way.

4.1 Structural Gamiﬁcation

At the ﬁrst stage of our application, we make use of

structural gamiﬁcation. As it is necessary, to train

Figure 5: Structural Gamiﬁcation Concept.

the classiﬁer (cf. Section 3.3) for activity and routine

recognition, manual inserts or corrections of activities

has been done for nearly two weeks. Furthermore,

a suitable amount of different sensors has also to be

present (especially Moves and Withings data).

The manual entering of these data and usage of ad-

ditional sensors has - at this point - no beneﬁt for the

user, as neither good classiﬁcation, nor healthy rou-

tine detection can be done. For that reason, we de-

cide to motivate the user with elements of structural

gamiﬁcation, as our preliminary study shows a strong

evidence between the raise of motivation and the el-

ements of “progress indication” and “levelling up”.

We introduce the elements experience points (XP) and

levels to our application, as these prove good results

for a short term motivation and also gives us a chance

not to overburden the user.

In the ﬁrst levels, the user has to earn XP by reg-

istering new sensors one after another. As we need

a base of different sensors for a good training set re-

sults, in the ﬁrst week, the user earns XP by gathering

sensor values for the Annotation component. This XP

are used to put it in certain levels that achieves the

next sensor and so on (“levelling up”).

In the process, we also show the current and

needed experience points to the user (“progress in-

dication”, cf. Figure 5). In that way, we can avoid to

confront the user with the internals of the classiﬁer or

a ﬁxed duration of training. By the use of the error

rate in activity recognition algorithm (cf. Sect. 3.3),

the system is able to unlock the Master Sensors and

the Advice Mode of the application (“content unlock-

ing”).

The structural gamiﬁcation stays always active,

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

238

even after all sensors are unlocked. Beside the experi-

ence points to unlock new sensors the user also earns

usable points with every achievements (“points”).

These points are used as a virtual currency to ensure

the system has always enough data to perform in a

stable way. If the usable point score falls under a

given threshold, the application becomes locked and

has to be reactivated by earning new points (i.e., by

re-activating sensors or training the classiﬁer).

4.2 Content Gamiﬁcation

At the second stage, the application uses elements of

content gamiﬁcation to establish a long-term motiva-

tion for the user. Based on the deviation detection, an

adaptive content gamiﬁcation can be applied.

The mentioned (cf. Section 3.5) proactive devia-

tion detection recognises anomalies in the daily rou-

tines, based on different time slots (e.g. week, month,

quarter). The system detects, if the routine differs

from previous ones and calculates on which basis this

difference was determined. For instance, if a daily

routine shows too short to Sleep cycles or unusual in-

termediate activities. The differing transition is used

for ﬁnding all causes in the sensor values.

These ﬁndings are used to build the dynamic gam-

iﬁcation elements. The system chooses the desired

feature (resp. sensor), e.g., mood or sleep duration,

and calculate the main factors inﬂuencing the chosen

feature (Franke et al., 2014c). For instance, features,

that affect the sleep duration could result in room

temperature, noise and sports duration. Beside the

inﬂuencing factors, we implemented rules that han-

dle whether this change is a positive or negative one.

The foundation of these assessments is derived from

chronobiology and formalised in our sensor module,

as previously stated (cf. Section 3.2).

The varied tasks are built due to this causes and

trends. If the sleep duration decreased signiﬁcantly

inﬂuenced by the sport duration, the next task will

be performing more sport, instead of confronting the

user with the real problem. This method avoids the

annoyance of the users, as they often recognise the

problems (no good sleep duration) themselves and

get frustrated if an application also shows the same

facts to him. Our study, e.g., shows that the With-

ings App showing such annoying messages: “You’ve

never done so few steps on a Wednesday. I’m sure you

can do more tomorrow :)”.

With this procedure, we cover the two signiﬁ-

cant content game elements “content unlocking” and

“varied challenges”, as the challenges change weekly

adapted to the current habits.

5 CONCLUSION

In this paper we presented a general gamiﬁcation con-

cept for daily routine deviations and gave an overview

of common used game elements in gamiﬁed apps for

self-improvement. In addition, we introduced our pre-

vious work SeMiWa in order to demonstrate and eval-

uate the suggested concept.

The purpose of our preliminary user study was to

identify game elements that have a particularly posi-

tive impact on the user’s motivation. We analysed var-

ious gamiﬁed apps for self-improvement and carved

out common used game elements that were used for

the online survey. Although the preliminary results

only consider a small number of participants, the out-

come shows, which game elements are already of im-

portance: progression, levels, challenge and curiosity.

Our future plans include completing the gamiﬁ-

cation concept and using it as a design framework for

gamiﬁed apps for self-improvement. A major require-

ment for our general concept is modularity. Game

elements and their visual representations are consid-

ered as two separate components. This is particularly

relevant regarding adaptivity, as it should be possi-

ble to vary game elements and visual representation

depending on the target group in the future. As a re-

sult, functionalities and the graphical user interface

will be adaptable according to user context (such as

age, gender, experience). In this regard, we will in-

clude a novel component into our concept to cover

the user context.

In order to be able to make a quantitative state-

ment on the motivational effect of the identiﬁed com-

mon used game elements, we will conduct an ex-

tended user study. The involvement of user context

aspects in our survey will allow for a more user-

centric gamiﬁcation concept. Identifying important

game elements in terms of various user context as-

pects, enables different rules for the Game Element

component as well as the Visual Game Element mod-

ule. This is necessary for reaching different target

groups.

Our study is also limited by its method, which

only provides an introduction of different apps for

self-improvement, not including the use of the apps

in real world context. The lack of interaction with the

gamiﬁed apps only enables replies to questions from

the user’s experience. For this reason we plan to fur-

ther develop our prototype, so that a qualitative user

study can be carried out in a real world context. Fur-

thermore, a long-term study with end users will al-

low us to investigate the game elements with regard

to their long-term motivation.

We believe our approach can offer added value for

AnAdaptive,StructuralandContentGamificationConceptforRegulatedDailyRoutines

239

giving awareness of regulated daily routines by pro-

viding a general gamiﬁcation concept as well as an

overview of signiﬁcant structural and content game

elements.

ACKNOWLEDGEMENTS

Part of this work has been executed under the project

VICCI (ESF-100098171) funded by the European

Social Fund and the German Federal State of Sax-

ony as well as within the project Dynapsys (BMWE-

19P12013D) by the German Federal Ministry for

Economic Affairs and Energy.

REFERENCES

Burmeister, D., Schrader, A., and Carlson, D. (2013). A

modular framework for ambient health monitoring. In

PervasiveHealth, pages 401–404. IEEE.

Chernbumroong, S., Cang, S., Atkins, A., and Yu, H.

(2013). Elderly activities recognition and classiﬁca-

tion for applications in assisted living. Expert Systems

with Applications, 40(5).

Deterding, S., Dixon, D., Khaled, R., and Nacke, L. (2011).

From game design elements to gamefulness: Deﬁning

”gamiﬁcation”. In Proceedings of the 15th Interna-

tional Academic MindTrek Conference: Envisioning

Future Media Environments, MindTrek ’11, pages 9–

15, New York, NY, USA. ACM.

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).

Franke, M., Keller, I., and Schlegel, T. (2014b). Mdro -

a modular daily routine ontology for assisted living.

In Workshop on Wearable Technology and Human -

Wearable Robot Interaction, (IEEE RO-MAN 2014).

Franke, M., Keller, I., and Schlegel, T. (forthcoming

2014a). Enabling cloud-based and personalized sen-

sors in assisted living environments. International

Journal on Collective Intelligence, 4(4).

Franke, M., Keller, I., and Schlegel, T. (forthcoming

2014c). Monitoring daily routine anomalies in as-

sisted living environments. In Workshop on Attention

for Social Intelligence, (ICSR 2014).

Franke, M., Lehmann, H., and Schlegel, T. (2014d). Cap-

turing and reusing empirical knowledge for activity

recognition in assisted living environments. In New

frontiers of service robotics for the elderly, (IEEE RO-

MAN 2014).

Franke, M., Seidl, C., and Schlegel, T. (2013a). A seamless

integration, semantic middleware for cyber-physical

systems. In Networking, Sensing and Control (IC-

NSC), 2013 10th IEEE International Conference on,

pages 627–632. IEEE.

Franke, M., Wuttig, A., and Schlegel, T. (2013b). Enabling

virtual, cloud-based sensors in assisted living environ-

ments. In 2013 IEEE 5th International Conference

on Cloud Computing Technology and Science (Cloud-

Com), volume 1.

Hamari, J., Koivisto, J., and Sarsa, H. (2014). Does gami-

ﬁcation work?–a literature review of empirical stud-

ies on gamiﬁcation. In System Sciences (HICSS),

2014 47th Hawaii International Conference on, pages

3025–3034. IEEE.

Hildebrandt, G., Moser, M., and Lehofer, M. (2013).

Chronobiologie und Chronomedizin Biologische

Rhythmen Medizinische Konsequenzen. Human

Research.

Kapp, K. M. (2013). The Gamiﬁcation of Learning and In-

struction Fieldbook: Ideas Into Practice. John Wiley

& Sons.

Kleinberger, T., Becker, M., Ras, E., Holzinger, A., and

Mller, P. (2007). Ambient intelligence in assisted

living: Enable elderly people to handle future in-

terfaces. In Stephanidis, C., editor, Universal Ac-

cess in Human-Computer Interaction. Ambient Inter-

action, number 4555 in Lecture Notes in Computer

Science, pages 103–112. Springer Berlin Heidelberg.

Malewski, C., Simonis, I., Terhorst, A., and Br

oring, A.

(2012). Starﬂ–a modularised metadata language for

sensor descriptions. International Journal of Digital

Earth, (ahead-of-print):1–20.

Minors, D. S. and Waterhouse, J. M. (2013). Circadian

rhythms and the human. Butterworth-Heinemann.

Sauermann, L., Van Elst, L., and Dengel, A. (2007). Pimo-

a framework for representing personal information

models. Proceedings of I-Semantics, 7:270–277.

Spork, P. (2011). Das Uhrwerk der Natur: Chronobiologie-

Leben mit der Zeit. Rowohlt Verlag GmbH.

Tan, R., Gu, J., Zhong, Z., and Chen, P. (2012). Socom:

Multi-sensor oriented context model based on ontolo-

gies. In Intelligent Environments (IE), 2012 8th Inter-

national Conference on, pages 236–242. IEEE.

Werbach, K. and Hunter, D. (2012). For the win: How

game thinking can revolutionize your business. Whar-

ton Digital Press.

Zulley, J. and Knab, B. (2013). Unsere innere Uhr:

nat

urliche Rhythmen nutzen und der Non-Stop-

Belastung entgehen. Mabuse-Verl.

BIODEVICES2015-InternationalConferenceonBiomedicalElectronicsandDevices

240