Mobile Crowd Sensing Services for Tinnitus Assessment, Therapy and Research [original]

Mobile Crowd Sensing Services for

Tinnitus Assessment, Therapy and Research

R¨

udiger Pryss, Manfred Reichert

Ulm University

Institute of Databases and Information Systems

ruediger[email protected],manfred.r[email protected]

Berthold Langguth, Winfried Schlee

University of Regensburg

Clinic and Policlinic for Psychiatry and Psychotherapy

[email protected],[email protected]

Abstract—Tinnitus, the phantom sensation of sound, is a

highly prevalent disorder that is difficult to treat; i.e., available

treatments are only effective for patient subgroups. Sufficiently

large and qualitative longitudinal data sets, which aggregate the

individuals’ demographic and clinical characteristics, together

with their response to specific therapeutic interventions, would

therefore facilitate evidence-based treatment suggestions for

individual patients. Currently, clinical trials are the standard

instrument for realizing evidence-based medicine. However,

the related information gathering is limited. For example,

clinical trials try to reduce the complexity of the individual

case by generating homogeneous groups to obtain significant

results. From the latter, individual treatment decisions are

inferred. A complementary approach would be to assess the

effect of specific interventions in large samples considering the

individual peculiarity of each subject. This allows providing

individualized treatment decisions. Recently, mobile crowd

sensing emerged as an approach for collecting large and

ecological valid datasets at rather low costs. By providing

mobile crowd sensing services to large numbers of patients,

large datasets can be gathered cheaply on a daily basis.

In the TrackYourTinnitus project, we implemented a mobile

crowd sensing platform to reveal new medical aspects on

tinnitus and its treatment. Additionally, we work on mobile

services exploring approaches for understanding tinnitus and

for improving its diagnostic and therapeutic management. We

present the TrackYourTinnitus platform as well as its goals,

architecture and preliminary results. Overall, the platform and

its mobile services offer promising perspectives for tinnitus

research and treatment.

Keywords-mobile crowd sensing, mobile healthcare applica-

tion, tinnitus, tinnitus variablity, clinical trial

I. INTRODUCTION

Tinnitus is a highly prevalent disorder (10-15 percent of

the population reports tinnitus) that currently has no suffi-

cient therapy [1]. Further, it is a purely subjective sensation

that can only be assessed by the report of the individual

patient. The pathophysiology of tinnitus is incompletely

understood and clinical trials frequently reveal contradictory

results. Presumably these non-conclusive results can be

explained by the fact that tinnitus is not a homogeneous

clinical entity. Instead, there exist many forms of tinnitus,

being distinct in their clinical characteristics as well as

response to specific therapeutic interventions [2]. Additional

complexity is introduced by the fact that the perception

of tinnitus loudness and distress is not constant in most

cases, but varies over time depending on the context (e.g.,

environmental sound level or stress) [3].

These inhomogeneous samples and the variability over

time provide an explanation for negative or non-replicable

findings encountered in most clinical tinnitus trials. Best

case, clinical trials can provide information on the efficacy

and safety of one therapeutic intervention in the investi-

gated sample. Furthermore, clinical trials generating such

data have been cost- and labour-intensive. In addition, the

procedure to involve and motivate patients is challenging and

the investigated patient sample is often not representative

due to restricted inclusion and exclusion criteria.

In order to mitigate these shortcomings, we developed

a mobile crowd sensing [4] platform called TrackYourTin-

nitus1(TYT). It tracks the individual tinnitus perception by

a specific questionnaire developed by us to assess tinnitus

perception and tinnitus-related parameters during the daily

routine of a patient. Additionally, the smart mobile device

of a patient records the environmental sound level while

the patient fills out the assessment questionnaire. Results

are transferred to the TYT backend, which, in turn, offers

features enabling researchers to evaluate gathered patient

data.

The remainder of this paper is organized as follows:

Section II introduces the TYT platform and its main features.

In Section III we discuss the current project status, whereas

Section IV presents project results. Section V discusses

mobile services built on top of the TYT platform. Section VI

discusses related work and Section VII concludes the paper

with a summary and outlook.

II. THE TRACKYOURTINNITUS MOBILE CROWD

SENSING PLATFORM

Tinnitus is a purely subjective phenomenon that is difficult

to measure. Moreover, tinnitus assessment is complicated by

the fact that tinnitus awareness and loudness vary over time,

1Further information can be found at: https://www.trackyourtinnitus.org

depending on current activities, environmental sound, stress

level, tiredness, and spontaneous fluctuations.

Magnetoencephalographic studies revealed that the mag-

nitude of functional connectivity between the brain areas

of the central auditory system and the ones responsible

for conscious perception, differs between tinnitus patients

and healthy controls [5], [6]. In turn, the intensity of this

connectivity correlates well with the tinnitus-related distress

reported by patients.

More recent research showed that the variability of os-

cillatory brain activity over time is reduced in the central

auditory system of tinnitus patients compared to controls

[7], which might influence the connectivity with the atten-

tional brain networks as well. Further research is needed to

evaluate in what way fluctuations of neuronal activity relate

to the variability of the subjective tinnitus perception. In

addition, for both diagnostic assessment of tinnitus patients

and outcome measurements of therapeutic interventions, an

exact assessment of an individual’s tinnitus is important.

However, in light of the variability of tinnitus loudness

and awareness under real life conditions, a comprehensive

assessment of tinnitus is challenging as well as cost- and

labour-intensive.

The TYT mobile crowd sensing platform aims at measur-

ing fluctuations of tinnitus perception and tinnitus distress

under real life conditions during a patient’s day as well

as for large numbers of patients. This allows tracking the

moment-to-moment fluctuation of the tinnitus. Furthermore,

tracked data may be related to everyday behavior and the

daily routine of patients to systematically identify relation-

ships between individual routines and tinnitus fluctuations.

Moreover, the TYT mobile crowd sensing platform can be

used to assess the effects of specific standardized therapeutic

interventions.

The TYT mobile crowd sensing platform has been devel-

oped in the context of a larger tinnitus database project2by a

multidisciplinary research team consisting of psychologists,

physicians and computer scientists. It comprises a website, a

backend and two mobile applications (cf. Fig. 1). The latter

track the individual tinnitus perception by providing three

core features:

1) Patients have to fill out a questionnaire (cf. Fig. 1 4

)

developed to assess tinnitus perception and tinnitus-

related parameters during the daily routine of a patient.

Thereby, patients are asked to complete the assessment

questionnaires at different times during the day on

a random basis (up to 12 notifications per day).

This procedure ensures that patients cannot foresee

the time of being asked and are involved in various

daily situations. Only when applying such randomized

approach, results might be of ecological validity.

2) In addition to the randomly applied questionnaire, for

2TINNET; http://tinnet.tinnitusresearch.net/

assessing momentary tinnitus loudness and distress,

once, users have to fill out three standardized tinnitus

questionnaires (cf. Fig. 1 3

) for the assessment of

stable tinnitus characteristics. Users may process them

with their smart mobile device or the website.

3) The smart mobile device records the environmental

sound level by using the integrated microphone, while

the patient fills out the assessment questionnaire.

Results are stored on the smart mobile device and

transferred to the TYT backend.

Several other aspects had to be considered when develop-

ing the apps. These aspects are either relevant for meeting

basic requirements of clinical practice (CP), for coping

with the technical environment (TI), or for increasing user

motivation (UM):

1) The questionnaires must run in the same way on all

supported mobile operating systems (CP).

2) User privacy must be ensured through secure data

transfer; produced data must be pseudonymized (CP).

3) It must be possible to build study groups (CP).

4) The TYT platform must ensure that the standardized

questionnaires are completed by the user before start-

ing the assessment based on the questionnaire. Note

that a user may enter the platform via the app (cf.

Fig. 1 1

) or the website (cf. Fig. 1 2

). Therefore, it

must be ensured that the standardized questionnaires

are completed in the same way using the app or the

website (cf. Fig. 1 3

) (CP).

5) The schema to randomly apply the assessment ques-

tionnaire to a patient must be stored locally on the

smart mobile device to be able to cope with long

periods of disconnection. In addition, patients must be

able to locally adapt the schema when the environment

changes (e.g., the user being on holidays; cf. Fig. 1 5

).

Further, the schema must be synchronized with the

TYT backend, and the feature to adapt the schema must

be provided in the same way on all mobile operating

systems and on the website (TI).

6) Processed assessment questionnaires and recorded

sound levels might produce large longitudinal data

sets. Data must be locally cached on the smart mobile

devices to cope with disconnections. Furthermore, it

must be securely transfered to the TYT backend to

prevent data loss as well as to ensure user privacy

(TI).

7) As an incentive, patients should be enabled to interact

with the TYT platform, e.g., to view the results of

the assessment questionnaires. This feature must be

provided on the smart mobile devices as well as on

the TYT website (UM).

Table I summarizes current features of the platform.

Sensor Integration

Backend with

Sensor

Framework

TrackYourTinnitus

REST-Interface

developed

planned

Oxygen

Saturation

Sensor

Fitness Tracker

Integration

Blood Pressure

Sensor

Heart Rate

Sensor

REST

REST Communication Style

Implemented Apps: iOS,

Android

App Store Deployments: 2

Downloads iOS: 1,045

Downloads Android: 673

Supported Languages:

German, English

Programming: Native Code

Developed Algorithms: 2

Developed Individual User

Controls:

9 per app

Used Frameworks:

AFNetworking, SIAlertView

App Development

Used Frameworks:

Twitter Bootstrap

Laravel

Mailchimp

Backend Development

Programming Language: PHP

Used Database: MySQL

Used Framework: Laravel

Programming Patterns: MVC

Used Protocols for Sensor

Integration: REST, Bluetooth

iOS

App

Android

App

REST REST

Windows

App

REST

Website Development

Sockets

Website Development

Bluetooth

Website Confirm E-Mail

Address Login to

Website

Fill out

Standardized

Questionnaires

Download

App

Website Confirm E-Mail

Address

Fill out

Standardized

Questionnaires

Fill out

Assessment

Questionnaire

Change

Notification

Schema

Use

Main Menu

Change

Notification

Ring Tone

View

Background

Information of

Project

View Results of

Assessment

Questionnaire

website

mobile apps

all standardized questionnaires completed

standardized

questionnaires

remaining

website account exisiting

no website

account

exisiting

all

standardized

questionnaires

completed

standardized

questionnaires

remaining

Figure 1: TrackYourTinnitus Platform

III. PROJECT STATUS

Table II presents current project figures (April 2015). The

project has been running for 12 months. We obtained 11,095

filled assessment questionnaires during this period, stem-

ming from more than 800 international users. The number of

users increases around 20 per week and hence, the number

of assessment questionnaires increases. In the beginning,

the TYT app and website were only provided in German

language. After three months, an English version was added.

Currently, we realize Spanish, French, Polish and Portuguese

versions. Psychometric validation of questionnaires in these

languages has shown that results are comparable [8].

We discuss some of the lessons learned made during the

project in more detail: First, we learned that, in general,

users are motivated to participate due to their health im-

pairment. However, when considering the figures presented

in Fig. 2, more incentives must be provided to increase

user motivation. Most of the randomly answered assessment

questionnaires were processed by only a small group of the

registered users. We investigated all gathered data of this

group and first results indicate that they suffer severely from

their tinnitus.

Hence, at this early stage, the developed mobile crowd

Feature Website iOS Android

Fill out standardized questionnaires √ √ √

Fill out assessment questionnaire −√ √

Visualize results √ √ √

Change notification schema √ √ √

Build study groups √ √ √

Table I: TrackYourTinnitus Features

sensing platform has primarily attracted severely affected

tinnitus patients. For motivating patients who are less

severely impaired, additional features are needed to increase

the overall benefit of the TYT app for patients. Currently,

the major added value of the TYT app for the patient is

the feedback on entered information. In order to increase

user motivation, we are developing a toolbox with different

features that may be helpful for reducing tinnitus perception

and annoyance. Examples of such features are auditory stim-

ulation, cognitive-behavioural therapy elements, social inter-

actions, and specific games. Another approach to address

user motivation will be to implement mechanisms enabling

users to register displeasure about existing TYT features.

Consequently, registered displeasure can be evaluated and

may be addressed.

Second, we are developing an additional questionnaire to

better understand why iOS is predominantly used.

Third, other research groups from the medical domain

Category Value

Project start 4/2014

Registered users 822

User home countries 75

Reported problems and failures 10

Number of developed questionnaires 4

Programmers 1

Team size 5

Emerged requests for using platform 5

APP downloads iOS 1,045

APP downloads Android 673

Processed assessment questionnaires 11,095

Processed standardized questionnaires 1,583

Totally gathered answers 90,343

Table II: TrackYourTinnitus Figures

answered Questionnaires

number of patients

050 100 150 200

12-10 11-50 51-100 >100

150 users processed

90% of all random

questionnaires

answered questionnaires

number of users

Figure 2: Assessment Questionnaires and User Activity

have encouraged us to realize features that allow customizing

the platform to specific needs. For example, to change the

questionnaires was often requested.

Fourth, we give insights into our expectations on the data

we want to collect with the platform in future. Tinnitus is

not the only prevalent disorder causing a large number of

severely impaired patients. In the future, the platform will

be applied in the context of other diseases as well. Its first

use in practice indicates that it is feasible in the healthcare

domain. In particular, it should be evolved to apply it in the

context of clinical trials with the goal to increase ecological

validity, while reducing costs at the same time. We expect

that the data collected with the TYT app will provide new

insights on the different subtypes of tinnitus.

Moreover, we expect that the amount of data collected

with the platform will significantly grow for two reasons.

First, we currently only provide German and English as

platform languages. As mentioned, other languages will

be added, which will result in a large number of addi-

tional users. Second, we are working on features that will

motivate more registered users to process the assessment

questionnaires. As shown in Fig. 2, 18% (150/822) of

the registerd users created the magnitude of the processed

questionnaires (90%). Furthermore, if other research groups

from the medical domain will largely collect data with the

TYT platform, a large multi-centric as well as multinational

data pool can be envisioned.

IV. PRELIMINARY RESULTS

This section presents preliminary results of the project.

First of all, the goals are discussed from a technical (T) as

well as a medical perspective (M) (cf. Table III). Then, the

achievements in respect to three of these goals are presented

in detail.

•T.Goal 1: An algorithm randomly notifying patients

was required to ensure ecological validity. In particular,

the algorithm behaves equally on all mobile operating

systems supported (i.e., iOS and Android)—we could

reach this goal by providing two different implemen-

tations to cope with the specific characteristics of the

respective mobile operating systems.

•T.Goal 2: An offline mode must be supported as well.

Consequently, data produced in offline mode must be

cached—such caching was implemented. However, to

also enable random notifications in offline mode, the

specific characteristics of the two mobile operating

systems need to be considered. While iOS offers a core

feature to implement respective notifications, Android

required us to implement it from scratch.

•T.Goal 5: A feature to view assessment results must be

provided—we evaluated various approaches to ensure

that user needs are met in the same way on both the

smart mobile devices and the website.

•T.Goal 6: A data export feature is required, which has

not been implemented yet. However, we add export

interfaces that will enable patients to interact with their

treating physician and allow clinicians to process data

with statistical software.

•M.Goal 3: In noisy environments, the tinnitus might

be partially or totally masked by surrounding sounds—

in the TYT app, background noise levels are recorded

in order to evaluate whether a reduction of tinnitus

awareness is caused by masking sounds or other factors.

•M.Goal 4: Users must get access to personal data to

learn more about their individual tinnitus. This will al-

low them to prevent behaviour worsening their tinnitus

and to deliberately engage in behaviour leading to an

improvement—we implemented respective features for

visualizing and displaying patient data.

•M.Goal 6: Users enter sensitive medical data with

the TYT app—to ensure privacy, all data gathered

are anonymized. Furthermore, users may delete their

account. Even if the account is deleted, data will be

kept at any time to ensure that the clinical trial will not

be manipulated—we implemented respective features

to ensure that all gathered data are anonymized and

clincial trials cannot be manipulated.

A. Notification Algorithm

We implemented an algorithm that applies the assessment

questionnaire to registered users on a random basis. As a

prerequisite, users have to specify a personal notification

Goals Description

Technical Goals

T.Goal 1 Develop notification algorithm.

T.Goal 2 Provide offline mode.

T.Goal 3 Provide similar mobile user interfaces.

T.Goal 4 Integrate website and apps properly.

T.Goal 5 Provide visualization of results.

T.Goal 6 Provide data export features.

Medical Goals

M.Goal 1 Collect longitudinal data for assessing individual tinnitus

fluctuation

M.Goal 2 Assess magnitude of tinnitus variability

M.Goal 3 Relate tinnitus perception to environmental noise

M.Goal 4 Provide feedback to patients

M.Goal 5 Evaluate crowd sensing for clinical trials

M.Goal 6 Ensure user privacy

Table III: TrackYourTinnitus Goals

schema when registering at the TYT platform (cf. Fig. 1 5

).

This schema comprises the following user-specified aspects:

First, the user must specify the number of notifications

applied on a daily basis. Second, users must specify the

days at which they want to be randomly notified; i.e., each

user must specify the time window he or she wants to be

randomly notified (e.g., Mondays between 2 and 6 p.m.).

The algorithm then uses the schema to calculate random

notifications for the respective user. Note that notifications

are realized based on the principle of local notifications;

i.e., they can be performed on smart mobile devices without

any connection to the TYT backend. Local notifications have

become necessary to be able to cope with longer periods of

disconnection. Due to the lack space, we omit details on how

we implemented local notifications on iOS and Android.

The schema is used by the notification algorithm as

follows:

1) The algorithm partitions the time window a user has

specified with respect to a particular day into ntime

intervals of equal length. ncorresponds to the number

of notifications the user has chosen.

2) The algorithm then calculates exactly one notification

for each interval. Thereby, it ensures that for each

notification the points in time for each notification are

randomly calculcated.

3) Finally, it is ensured that there are at least 15 minutes

between two notifications.

We only present the algorithm running on iOS (cf. Algo-

rithm 1) and the calculated notifications for a single day. In

practice, notifications are calculated in advance.

Algorithm 1: iOS algorithm for daily notifications of a user

Data:

timeInterval: time interval a user has specified for a day

numberOfNotificationsP erDay: notifications specified for a day

Result:

scheduleLocalNotification: calculated random notifications for a day

1begin

2lengthOfIntervall = timeInterval/numberOfNotificationsPerDay;

3lastNotification = 900; /*the 15 minutes */

4foreach n∈numberOfNotificationsP erDay do

5secondsSinceStartOfInterval =

arc4random uniform3(lengthOfIntervall);

6absoluteInterval=

7secondsSinceStartOfInterval+(n*lengthOfIntervall);

/*check the 15 minutes */

8if absoluteInterval −lastNotification < 900 then

9absoluteInterval = 2*absoluteInterval - lastNotification;

10 end

11 lastNotification = absoluteInterval;

/*check if notification is in

absoluteInterval */

12 if absoluteInterval < timeInterval then

/*notification found */

13 scheduleLocalNotification =

scheduleLocalNotification ∪absoluteInterval;

14 end

15 end

16 end

3arc4random uniform(upper bound): iOS internal function to return a

uniformly distributed random number less than upper bound.

X X X X

n=4, day=Monday

2 p.m. 6 p.m.

>= 15 minutes

60 minutes per interval

Figure 3: Example of Algorithm 1

Consider Line 12 of Algorithm 1. It may happen that

a user is notified after the end of the time window spec-

ified by the user. These notifications are not considered

for the scheduleLocalNotification of a day and hence

reduce numberOfNotificationsPerDay. The approach

has proven its feasbility for practical as well as statistical

use. Fig. 3 presents a computation example for Monday with

a user-specified time window between 2 and 6 pm.

Altogether, we have not changed the algorithm since

project start (4/2014). It has worked properly from a tech-

nical perspective (i.e., no problems were reported by TYT

users). From a statistical perspective, more data is needed

to fully evaluate the appropriateness of the algorithm in the

large scale.

B. Assessment of the magnitude of tinnitus variability

Figures 4-6 present clinical data of individual patients we

gathered with the TYT platform to assess and investigate the

magnitude of tinnitus variability.

Fig. 4 shows data of a tinnitus patient with a large

variability of the tinnitus loudness. The patient has answered

almost 400 notifications using the mobile app. The variation

of tinnitus loudness is shown on the ordinate.

0100 200 300

0.0 0.2 0.4 0.6 0.8 1.0

sampling points

Tinnitus loudness

sampling points

tinnitus loudness

sampling points

Figure 4: Tinnitus Perception and Large Variability

Fig. 5 shows data of a tinnitus patient with a strong rela-

tionship between tinnitus perception and the environmental

sound level that was measured by the mobile TYT app when

the patient was answering the assessment questionnaire.

The measurements of the sound pressure level have been

normalized (z-transformation). In quiet environments, the

tinnitus loudness varied between 0.1 and 0.7. In turn, in

loud environments the tinnitus was always suppressed to a

level below 0.2.

0123

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Environmental Noise

Tinnitus Loudness

Normalized

sound level

normalized sound level

tinnitus loudness

Figure 5: Tinnitus Perception and Environmental Sound

Level

Finally, Fig. 6 shows data of a tinnitus patient with a clear

relationship between the subjective perception of tinnitus

and the time of day. The tinnitus loudness ratings were

averaged for the hours from 8 am to 11 pm. In the morning,

the patient perceives the tinnitus with reduced loudness.

During the day, the perceived loudness of tinnitus increases

up to its maximum at night.

Time of day

0.2 0.3 0.4 0.5 0.6 0.7 0.8

time of day

Tinnitus Loudness

8:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00

time of day

tinnitus loudness

8am

10am

12pm

2pm

4pm

6pm

8pm

10pm

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Figure 6: Tinnitus Perception and Time of Day

Altogether, first results have shown that the magnitude

of the perceived tinnitus loudness can vary largely. This

variation can be related to the sound pressure level of

the surrounding environmental sounds or to the time of

day. Other factors are currently under investigation. This

variation of tinnitus perception may represent an important

confounding factor for clinical trials.

C. Evaluation of mobile crowd sensing for clinical trials

Today, clinical trials usually measure the tinnitus loud-

ness at one point in time before the start of the clinical

intervention and at one point in time directly after finishing

it. This routine, however, does not consider the variations

of tinnitus loudness as measured with the TYT platform. In

fact, the variation of tinnitus loudness can introduce a large

variance in the data of the clinical trial that is not related

to the clinical intervention per se. Based on this data, we

suggest refining the standard protocols of clinical trials in

the field of tinnitus by adding additional measurement points

before and after the intervention for a better estimation of the

true effect introduced by the clinical intervention. Note that

the described mobile application enables such refinements.

V. FURTHER MOBILE CROWD SENSING SERVICES

This section presents further mobile services related to

the TYT platform (cf. Fig. 7). Their development has been

driven by findings obtained when running the project over

12 months. Note that our vision is to utilize the findings

of the TYT platform for enabling new diagnostic and ther-

apeutic approaches. At the current project stage, we have

already prototypically implemented the Tinnitus Navigator

app, whereas the TYT Feedback app is in planning stage.

Additionally, the TYT platform has been extended taking

the gathered findings into account.

A. Tinnitus Navigator

Tinnitus Navigator is realized as a mobile application and

will be connected to the same website as the TYT apps. The

Tinnitus Navigator aims to assist treating physicians in the

diagnostic and therapeutic management of a tinnitus patient.

In particular, it will provide treatment suggestions based on

a patient’s individual clinical profile. Treatment suggestions,

in turn, will be based on a growing database that incorporates

treatment guidelines, data from clinical trials, and longitudi-

nal data from the TYT mobile crowd sensing platform. Rec-

ommendations are continuously updated through feedback

from the Tinnitus Navigator. This mechanism ensures that

recommendations, which do not provide the expected results,

are continuously refined. Currently, the first prototypes of the

Tinnitus Navigator mobile app on Android and on iOS have

been implemented to address interface requirements.

B. TrackYourTinnitus Extensions

Two additional features (cf. Fig. 7 1

,2

) were developed

for the TYT platform. They were motivated by user requests

running the project. First, we developed a mobile service en-

abling patients to determine the individual tinnitus frequency

on their own (cf. Fig. 7 1

). Utilizing this information,

patients can establish a therapy with the practitioner or adjust

a running one.

We integrated three sensors as shown in Fig. 1. Previous

work suggests that the conscious perception of the phantom

REST

Future

Objective

Implemented Apps:

Android, iOS

App Store Deployments: 0

Languages Supported:

German, English

Programming: Native Code

Developed Algorithms: 2

Developed Individual User

Controls: 9

iOS

App

Android

App

Backend with Sensor

Framework

Website

TrackYourTinnitus and Extensions

REST-Interface

REST

Sockets

Research and Development Timeline t

Goals:

(1) Provide personal data

to obtain personal

treatment suggestions

(2) Manage personal

tinnitus record

(3) Share information with

other users

(4) Prepare data for

treating doctor

Tinnitus Navigator

Tinnitus Research Data Pool (2) fosters development of

(1) Integration of Social

Networks, e.g.,

REST

(2) Provide In-App

Tinnitus User Forum

TrackYourTinnitus Patient Feedback

(3) envisions development of

(1) allows for

REST

developed

planned

REST Communication Style

Extension 1:

Determine personal

tinnitus frequency

with apps

and website

Extension 2:

Determine vital

signs with apps

Extension 3:

Provide user feedback based on

new algorithms

Figure 7: Mobile Services for Tinnitus Research and Treatment

tinnitus sound depends on more parameters than the recorded

sound level. Others might be medication, emotional arousal,

stress, alcohol, caffeine consumption, infections, hormone

levels, rural versus urban environment, sleep quality, circa-

dian and circaannual rhythm, or comorbidities. In order to

collect more relevant contextual information, the three sen-

sors were integrated to gather additional relevant data such

as oxygen saturation or cardiac frequency. Thus, the mobile

crowd sensing technology enables a detailed assessment of

these parameters on tinnitus and annoyance. Note that a

recent study has revealed the usefulness of large datasets for

elucidating such relationships. The study analyzed Internet

search engine query data to identify seasonal trends in

tinnitus severity [9].

Currently, we are developing algorithms to automatically

evaluate gathered patient data (cf. Fig. 7 3

. Either these

algorithms calculate individual therapy suggestions for a pa-

tient or trigger other components being able to automatically

refine therapy suggestions. Altogether, first experiments we

made with the TYT platform revealed that intelligent feed-

back on collected data is essential for increasing the patient

motivation to use the app.

VI. RELATED WORK

Different categories of related work are relevant in the

given context:

Approaches dealing with mobile crowd sensing [4], [10]–

[12].First, there are approaches that develop program-

ming frameworks enabling users to easily configure mobile

crowd sensing applications. For example, the framework

presented in [12] enables users to configure such applications

based on tasks, which can be specified in a high-level and

user-friendly notation. To realize a collaborative learning

application, tasks Recruit, GetRawData, GetFeatures, and

UploadFeatures must be specified.

Second, there are approaches dealing with a specific

mobile crowd sensing application scenario. For example,

[11] utilizes Twitter for its mobile crowd sensing application.

One of the applications presented in [11] evaluates recorded

noise levels with the help of Twitter information. Thereby,

smart mobile devices of many users automatically determine

the local sound level and transfer recorded data to the

Twitter platform. With this information, for example, it may

be determined for a particular location whether a party

is currently taking place. Third, there are approaches that

investigate for which application scenarios mobile crowd

sensing is useful [12].

Approaches utilizing mobile crowd sensing technology for

clinical or psychological trials. Interestingly, mobile crowd

sensing technology is still rarely used in a clinical context.

This may be related to legal and data privacy issues [13],

but also to a general resistance of health systems to adopt

innovative data information technologies. Today, still the

magnitude of clinical data is paper-based. However, it is

expected that mobile and big data technologies [14], [15],

with their potential to revolutionize clinical research and

clinical trials, will enter the medical field.

Approaches that deal with mobile data collection based on

psychological and clinical questionnaires. Recently, various

mobile applications have been developed for psychological

studies [16], [17]. In order to fully capitalize their potential,

the pure adoption of existing questionnaires for mobile use

will be outperformed by novel concepts for information

collection [18], [19].

In summary, in many different life domains the feasibility

of mobile crowd sensing has been already proven. The med-

ical field, albeit a theoretically highly promising application

for crowd sensing approaches, seems to be still neglected.

VII. OUTLOOK AND SUMMARY

This paper introduced the TYT mobile crowd sensing plat-

form. We presented the current status of its implementation

and practical use. Furthermore, we discussed preliminary

results we obtained when running the platform for over 12

months. In particular, we showed that these results indicate

new insights on the tinnitus variability. We further showed

that the obtained results provide the basis to develop new

mobile crowd sensing services fostering tinnitus assessment,

therapy and research. Moreover, the results indicate that

users are actually motivated to use the platform, especially

those severely suffering from tinnitus. Still more incentives

and features are required to increase user motivation and

hence to gather more valuable data on the tinnitus disease.

Therefore, we are working on algorithms to automatically

evaluate patient data in order to provide immediate valuable

feedback to them. Altogether, using mobile crowd sensing

and its application offers promising perspectives for tinnitus

assessment, therapy and research as well as for the medical

field in general.

REFERENCES

[1] B. Langguth, “A review of tinnitus symptoms beyond’ringing

in the ears’: a call to action,” Current Medical Research &

Opinion, vol. 27, no. 8, pp. 1635–1643, 2011.

[2] M. Landgrebe, F. Zeman, M. Koller, Y. Eberl, M. Mohr,

J. Reiter, S. Staudinger, G. Hajak, and B. Langguth, “The

tinnitus research initiative (tri) database: a new approach for

delineation of tinnitus subtypes and generation of predictors

for treatment outcome,” BMC medical informatics and deci-

sion making, vol. 10, no. 1, p. 42, 2010.

[3] W. Schlee, J. Herrmann, R. Pryss, M. Reichert, and

B. Langguth, “How dynamic is the continuous tinnitus per-

cept?” in 11th Int’l Tinnitus Seminar, May 2014.

[4] N. Lane, E. Miluzzo, H. Lu, D. Peebles, T. Choudhury, and

A. Campbell, “A survey of mobile phone sensing,” IEEE

Communications Magazine, vol. 48, no. 9, pp. 140–150, 2010.

[5] W. Schlee, N. Mueller, T. Hartmann, J. Keil, I. Lorenz, and

N. Weisz, “Mapping cortical hubs in tinnitus,” BMC biology,

vol. 7, no. 1, p. 80, 2009.

[6] W. Schlee, T. Hartmann, B. Langguth, and N. Weisz, “Abnor-

mal resting-state cortical coupling in chronic tinnitus,” BMC

neuroscience, vol. 10, no. 1, p. 11, 2009.

[7] W. Schlee, M. Schecklmann, A. Lehner, P. Kreuzer,

V. Vielsmeier, T. Poeppl, and B. Langguth, “Reduced Vari-

ability of Auditory Alpha Activity in Chronic Tinnitus,”

Neural Plasticity, vol. 2014, 2014.

[8] D. Zeman, “Data issues of the multilingual translation ma-

trix,” in Proc of the Seventh Workshop on Statistical Machine

Translation. Association for Computational Linguistics,

2012, pp. 395–400.

[9] D. Plante and D. Ingram, “Seasonal trends in tinnitus symp-

tomatology: evidence from Internet search engine query data,”

European Archives of Oto-Rhino-Laryngology, pp. 1–7, 2014.

[10] R. Ganti, F. Ye, and H. Lei, “Mobile crowdsensing: cur-

rent state and future challenges,” Communications Magazine,

IEEE, vol. 49, no. 11, pp. 32–39, 2011.

[11] M. Demirbas, M. Bayir, C. Akcora, S. Yilmaz, and H. Fer-

hatosmanoglu, “Crowd-sourced sensing and collaboration us-

ing Twitter,” in 2010 IEEE Int’l Symp on a World of Wireless

Mobile and Multimedia Networks (WoWMoM). IEEE, 2010,

pp. 1–9.

[12] M. Ra, B. Liu, T. La Porta, and R. Govindan, “Medusa: A

programming framework for crowd-sensing applications,” in

Proc of the 10th Int’l Conf on Mobile systems, applications,

and services. ACM, 2012, pp. 337–350.

[13] D. Christin, A. Reinhardt, S. Kanhere, and M. Hollick, “A

survey on privacy in mobile participatory sensing applica-

tions,” Journal of Systems and Software, vol. 84, no. 11, pp.

1928–1946, 2011.

[14] J. Laurila, D. Gatica-Perez, I. Aad, J. Blom, O. Bornet,

T. Do, O. Dousse, J. Eberle, and M. Miettinen, “The mobile

data challenge: Big data for mobile computing research,” in

Pervasive Computing, no. EPFL-CONF-192489, 2012.

[15] R. Pryss, N. Mundbrod, D. Langer, and M. Reichert, “Sup-

porting medical ward rounds through mobile task and process

management,” Information Systems and e-Business Manage-

ment, vol. 13, no. 1, pp. 107–146, 2015.

[16] J. Schobel, R. Pryss, and M. Reichert, “Using smart mobile

devices for collecting structured data in clinical trials: Results

from a large-scale case study,” in IEEE 28th Int’l Symposium

on Computer-Based Medical Systems (CBMS), 2015.

[17] A. Crombach, C. Nandi, M. Bambonye, M. Liebrecht,

R. Pryss, M. Reichert, T. Elbert, and R. Weierstall, “Screening

for mental disorders in post-conflict regions using computer

apps - a feasibility study from burundi,” in XIII Congress

of European Society of Traumatic Stress Studies (ESTSS)

Conference, June 2013, pp. 70–70.

[18] J. Schobel, M. Schickler, R. Pryss, F. Maier, and M. Reichert,

“Towards Process-Driven Mobile Data Collection Applica-

tions: Requirements, Challenges, Lessons Learned,” in 10th

Int’l Conf on Web Information Systems and Technologies,

April 2014, pp. 371–382.

[19] J. Schobel, M. Ruf-Leuschner, R. Pryss, M. Reichert,

M. Schickler, M. Schauer, R. Weierstall, D. Isele, C. Nandi,

and T. Elbert, “A generic questionnaire framework supporting

psychological studies with smartphone technologies,” in XIII

Congress of European Society of Traumatic Stress Studies

(ESTSS) Conference, June 2013, pp. 69–69.