MoBio

A Mobile Application for Collecting Data from Sensors

Petr Je

zek and Roman Mou

cek

New Technologies for the Information Society, Faculty of Applied Sciences, University of West Bohemia,

Univerzitn

ı 8, 306 14 Plze

n, Czech Republic

Keywords:

MoBio, Sensor, Android, Mobile Application, Brain Data, Health Data, Domain Terminology, Data Transfer,

EEGBase.

Abstract:

There are a lot of sensors for monitoring human health and/or ﬁtness level on the market. They facilitate

collection of data from the human body and advanced devices even facilitate data transfer to remote servers

where the collected data are further processed. While health data, obtained e.g. from accelerometers or chest

straps, are collected rather frequently, brain electrophysiology data, obtained from surface electrodes, are

still collected relatively rarely. However, integration and correlation of brain signals with other sensory data

would be very interesting for next research of physical and mental health. Although capturing brain signals

in real environment still faces technological difﬁculties, current development of common infrastructure seems

to be useful. Then this article deals with various architectures and data formats used for storage and transfer

of sensory data and their possible integration with existing neuroinformatics approaches. As a solution we

introduced a terminology describing data from a limited collection of sensors. The terminology is implemented

in the odML format and integrated in a proof-of-concept Android application. Data transfer, storage and

visualisation as well as integration with a remote neuroinformatics resource are presented.

1 INTRODUCTION

There are a lot of factors affecting human health.

Some of them such as genetics, environmental inﬂu-

ence or internal state of an individual cannot be eas-

ily measured. On the other hand, there are the fac-

tors such as blood pressure, glucose level or heart

rate that can be measured relatively easily and non-

invasively using cheap sensors. For a long time elec-

trophysiological measurements have been conducted

in laboratories equipped by common desktop comput-

ers and non-transferable measuring devices. Fortu-

nately, this situation has been rapidly changing. One

of the reasons is increasing popularity of smart de-

vices such as smart-phones or tablets. According to

eMarketer (eMarketer, 2015) two billion people will

own a smart-phone in 2016. Simultaneously, a lot

of relatively cheap sensors for measuring potentials

from the human body are available on the market.

The data produced by these sensors are usually trans-

ferred wirelessly, consequently they can be read and

processed by smart devices. This technical progress

enables a particular shift of treatment from hospitals

to home environments and facilitates collecting data

during outdoor activities. The obtained data can be

used in two fundamental intersecting ways.

At ﬁrst, assistive technology approach serves to

stimulate, maintain, and improve functional capabili-

ties of people with special needs including disabled

people or aging population. Getting independence

and self-sufﬁciency increases the quality of life in

general.

The second approach is focused on sportsmen or

actively living people. People are being monitored

when they are performing speciﬁc activities (e.g. run-

ning or long distance walking). The data are used for

a long term monitoring of their ﬁtness level.

So far we have not talked about a signiﬁcant

source of electrophysiology data, the human brain.

Although there are technological improvements for

capturing surface brain data in real environment, this

kind of data collection is still not widespread. How-

ever, integration and correlation of brain signals with

other sensory data would be very interesting for next

research of physical and mental health.

Our research group operates a completely

equipped laboratory (Moucek et al., 2014) for electro-

physiological measurements. We are focused mainly

on experimental work using the methods and tech-

niques of electroencephalography (EEG) and event-

Ježek, P. and Mou

cek, R.

MoBio - A Mobile Application for Collecting Data from Sensors.

In Proceedings of the International Conference on Information and Communication Technologies for Ageing Well and e-Health (ICT4AWE 2016), pages 115-121

ISBN: 978-989-758-180-9

115

related potentials (ERP). Except of the permanent lab-

oratory we also operate a mobile laboratory equipped

by a set of laptops and portable measuring devices

for performing experiments in the external environ-

ment. With advancing efforts to extend the laboratory

to collect diverse collections of data (e.g. blood pres-

sure, glucose level, heart rate) we are extending our

infrastructure (hardware devices and software tools)

to support measurements of heterogeneous data from

various kinds of sensors.

In this paper we deal with various architectures

and data formats used for storage and transfer of sen-

sory data, and their possible integration with existing

neuroinformatics approaches. Since data from exist-

ing sensors are usually stored in proprietary formats

and transferred to closed databases, it is difﬁcult to

use them in experimental laboratories. As a solu-

tion we introduce a terminology describing data from

a limited collection of sensors. Then a prototype of a

mobile client for collecting data from these sensors is

presented. This client provides integration with a lim-

ited set of devices and enables data to be transferred

and visualized in a remote storage.

2 STATE OF THE ART

According to (Lowe and Laighin, 2012) applications

using sensors can be divided into three categories.

The ﬁrst category, Smart Phone Applications, use ei-

ther GPS or on-board kinematic sensors as the tech-

nologies of choice for monitoring exercise. The sec-

ond category comprises of any system that uses a cen-

tral controller and an external sensor. The last cate-

gory comes from image processing domain. It uses a

combination of a computer screen and a camera. The

camera monitors the exact movement and position of

the entire body during exercise. The screen is used for

the interaction with the user.

The ﬁrst category represents applications such as

Endomondo or Runkeeper. The second category is

represented by e.g. Nike+, miCoach, Garmin Heart

Belt or Fora Active tonometer. A typical representa-

tive of the third category is Microsoft Kinect. While

Microsoft Kinect is designed for the indoor use, all

other devices are designed mostly for the outdoor use.

Available tools and sensors are usually designed

for one speciﬁc activity. They are able to record only

limited variety of data. In addition, due to their pro-

prietary structure they do not provide suitable inter-

faces for integration with other systems. Issues with

deloying commercial biosensors for a body sensor

network are described in (Seeger et al., 2011).

As well as other communities neuroinformatics

community identiﬁed problems with a long-term de-

scription, storage and management of experimental

data/metadata (Teeters et al., 2008). To facilitate solu-

tion of these difﬁculties International Neuroinformat-

ics Coordinating Facility (INCF) (INCF, 2013) covers

activities for development of infrastructures and data

standards in neuroinformatics community. Ontology

for Experimental Neurophysiology (OEN) (Le Franc

et al., 2014a) uses semantic web approach to describe

terminology used in biomedical science and neuro-

science. Our research groups developed a system for

long term storage and management of EEG/ERP ex-

perimental data and metadata, EEGBase (Jezek and

Moucek, 2012) that was recently extended by a sys-

tem of templates suitable for storing various data and

metadata structures. A mobile EEGBase client (Jezek

and Moucek, 2013) is a supplementary Android tool

that enables collecting experiments out of the labo-

ratory and provides an on-line synchronization with

EEGBase.

3 SENSOR ARCHITECTURES

AND DATA TRANSFER

Sensors use different ways to manipulate data. Some

sensors have their own displays or internal memories

for storing measured data that can be later transferred

to a long term storage. These devices usually provide

a complete history (of course, limited by the capacity

of their internal memories) of stored measurements.

On the other hand, devices such as heart rate meters

do not have any internal memory. Data can be trans-

ferred only at the time of measurement. According

to the way the data are distributed, we deﬁned three

levels of sensor architecture.

• one-layer - The sensor is a stand-alone unit having

a display, memory, and a set of controls. Such a

sensor is completely self-controlled and does not

establish any connection to a remote system.

• two-layers - The sensor captures data that are ﬁ-

nally sent to a remote system (the sensor may/may

not have an internal memory). This remote sys-

tem can be a common computer/laptop or a smart

phone/watch etc.

• three-layers - A cloud service is added to the two-

layer architecture. The captured data are sent to

a remote server where they are stored, managed,

and analyzed.

A lot of sensors belonging to the one-layer archi-

tecture, including various glucometers, tonometers,

and step counters, store captured data in a proprietary

ICT4AWE 2016 - 2nd International Conference on Information and Communication Technologies for Ageing Well and e-Health

116

Table 1: Tested e-health and ﬁtness systems.

Intern.

sensors

Extern.

sensors

Conti-

nuous

measure-

ments

Single

measure-

ments

Stats

Cloud

Service

Design

Endomondo

www.endomondo.com

2 3 • • • +

Runstatic

www.runtastic.com

2 2 • • • +

Pedometer

www.runtastic.com

/en/apps/pedometer

1 • • • -

Push-ups

www.runtastic.com/

en/apps/pushups

1 • • • +

Heart rate

www.runtastic.com

/en/apps/heartrate

1 • • • +

Sport Tracker

www.sports-tracker.com

1 2 • • • +

Health Tracker

play.google.com/store/apps/

details?id=com.benoved.phr lite

• -

Madbarz

madbarz.com

• • +

eVito

www.evito.cz

1 5 • • • • +

storage. These devices usually do not require/provide

any connection to other systems. Representatives of

two-layer architecture include sensors such as heart

rate belts or step counters. They often do not have

their own display and usually use Bluetooth or ANT

(resp ANT+)

technologies for data transfer. Data are

usually displayed on a screen of a smart device. Be-

cause of the limited performance of mobile devices

only a basic data processing is provided. Advanced

data processing is performed on a remote server that

forms part of the three-layer architecture. There are

few exceptions such as BlibCare tonometer equipped

with WiFi. Then the captured data can be transferred

directly by using e.g. a home WiFi network.

4 TESTED SYSTEMS

E-health and ﬁtness systems that use the three-layer

sensor architecture described above are most suit-

able for long term storage and management of col-

lected data. When providing a remote data transfer

ANT is the protocol while ANT+ is a set of mutually

agreed deﬁnitions what the information sent over ANT rep-

resents.

to a server, these devices are also suitable to be in-

tegrated with neuroinformatics infrastructures. The

biggest obstacle encountered is usually a data format

of transferred data. When devices implement Blue-

tooth or ANT+ technologies, they can use a lot of de-

ﬁned proﬁles including Blood Pressure Proﬁle (BLP),

Heart Rate Proﬁle (HRP), Health Thermometer Pro-

ﬁle (HTP)

for Bluetooth, or Weight Scale and Heart

rate

for ANT+.

Table 1 summarizes and provides evaluation of the

e-health and ﬁtness systems we tested. Only the sys-

tems providing at least a cloud service as a storage or

statistical results as an analytic result were selected.

For all systems we considered the number of sup-

ported internal (e.g. a GPS sensor on a smart phone)

and external sensors. This number is very limited as

can be seen in Table 1. Two systems do not support

any sensor; data are inserted manually. The last col-

umn Design is a subjective evaluation of the user in-

terface. Although all tested systems use the ANT or

Bluetooth technology, they implement a proprietary

transfer format instead of using any available proﬁle.

https://developer.bluetooth.org/TechnologyOverview/

Pages/Proﬁles.aspx

https://www.thisisant.com/developer/ant-plus/

device-proﬁles

MoBio - A Mobile Application for Collecting Data from Sensors

117

5 MOBILE APPLICATION

PROTOTYPE

5.1 System Scope

Having difﬁculties with closed-source dedicated de-

vices we present a prototype of a mobile application

that solves the issues described above. It aggregates

data from various sensors into a ﬂexible data storage

on the server. The data can be used by human readers

or processed by automatic readers. The application

can be easily used outdoors and in areas without In-

ternet connectivity. When the client gets on-line, the

stored data are synchronized with a remote storage.

However, the main core of this solution is a proposal

and implementation of suitable terminology for sen-

sory data.

5.2 Format Selection

Since the aim of the application is to support large

collections of sensors, it must store data in a ﬂexible

data format. Existing data formats are based on differ-

ent levels of abstraction and include low-level binary

formats, highly abstract implementation-independent

data formats, and formats based on ontologies. We

required a format that provided a sufﬁcient level of

abstraction to be system independent, but also easy-

to-use format having no speciﬁc demands on users.

Having been involved in neuroinformatics commu-

nity and having been persuaded that the similar so-

lution could be used for other sensory data, we pre-

ferred open-source formats supported and accepted in

this domain.

The working group of the INCF Task Force on

Electrophysiology

introduced two approaches to-

wards deﬁning a standard, which may eventually

be merged (Teeters et al., 2013). The ﬁrst one

uses the Hierarchical Data Format (HDF5) (HDF5

Group, 2013) or respectively epHDF, the specialized

HDF5 format for electrophysiology. The NIX for-

mat (Stoewer et al., 2014) provides a data model

for storing experimental data in HDF5, together with

metadata in the odML format (Grewe et al., 2011).

We selected odML (odML is a free form tree-like

structure of sections, properties and values) as a suit-

able format for the presented application because of

its platform-independence, simplicity, and human-

readability. Moreover, it ensures a compatibility with

other systems developed in neuroinformatics commu-

nity, for example (Zehl et al., 2014), (Le Franc et al.,

http://www.incf.org/programs/datasharing/

electrophysiology-task-force

Figure 1: Proposed Terminology.

2014b), and (Davison et al., 2013).

5.3 Terminology

Figure 1 shows a terminology based on the odML for-

mat and proposed for measurements of blood pres-

sure, heart rate, glucose level, and weight scale in-

cluding hydratation, muscle mass or bone mass. List-

ing 1 shows a JSON representation of the blood

pressure data structure, the section blood presure

with properties systolic pressure, diastolic pressure,

mean pressure, and a subsection heart rate are de-

ﬁned. The document is simpliﬁed to keep readabil-

ity. A complete document contains a complete set of

properties describing the experiment. Because of the

ﬂexibility of the odML format, this terminology can

be easily extended for other measurements.

Listing 1: Blood pressure example.

{

" me tada t a " :{

" o dM L " :{

" d at e " : " 2 0 1 5 - 0 7 - 0 3 " ,

" x m ln s : g ui " : " ht tp : / / www . g - nod e . or g /

gu im l " ,

" s e c ti o n " : [

{

" n am e " : " b loo d _ p r e ss u r e " ,

" pr o per t y " : [

{

" n am e " : " sy s t ol i c " ,

" v a lu e " :{

" t yp e " : " i nt " ,

" c o n te n t " : 8 0

{

" n am e " : " dya s t ol i c " ,

" v a lu e " :{

" t yp e " : " i nt " ,

" c o n te n t " : 6 0

ICT4AWE 2016 - 2nd International Conference on Information and Communication Technologies for Ageing Well and e-Health

118

{

" n am e " : " m ean " ,

" v a lu e " :{

" t yp e " : " i nt " ,

" c o n te n t " : 7 0

" t yp e " : " b loo d _ p r e ss u r e "

}

" s e c ti o n " : [

{

" n am e " : " h ea r t _ ra t e " ,

" l in k " : " b loo d _ p r e ss u r e " ,

" pr o per t y " : [

{

" n am e " : " b ea t _ c ou n t " ,

" v a lu e " :{

" t yp e " : " i nt " ,

" c o n te n t " : 5 2

{

" n am e " : " bea t _ ti m e " ,

" v a lu e " :{

" t yp e " : " i nt " ,

" c o n te n t " : 1

" t yp e " : " h ea r t _ ra t e "

}

" v e r si o n " : 1

}

5.4 Architecture and Implementation

To support the use of open source technologies we

implemented the MoBio application for the Android

platform that is available on a wide range of devices

including tablets and mobile phones (according to

StatCounter

more than 60% of devices are operated

by Android). Moreover, there are other advantages to

use this platform, for example a lot of cheap devices

on the market, large community of developers, and

easy publication of applications.

Figure 2 shows the used architecture. The user

collects data using a body sensor. The data are trans-

ferred to MoBio running on a smart phone using

an ANT+ or Bluetooth proﬁle. Then MoBio transfers

data expressed in an odML terminology to the server.

The presented proof of concept implementation

provides basic functionality including management of

http://gs.statcounter.com

Figure 2: MoBio System Architecture.

user accounts, support for a limited set of sensors, and

visualization of stored data. Collected data are stored

on a SD card and can be visualized directly on the

mobile phone screen or transferred to the server and

stored there for future processing. The registration

form includes both basic information about the user

such as his/her name or email and advanced infor-

mation such as gender, weight, height, current ﬁtness

level, etc. All these data are required because they can

affect results of measurements.

Supported devices must enable either Bluetooth

or ANT transfer. In the current implementation we

successfully tested a limited set of devices produced

by Garmin and Fora producers. When the device is

paired, the user can transfer data. Figure 3 shows a

basic functionality provided to the logged user. Fig-

ure 3a shows a list with a paired device, the Garmin

Heart Rate Sensor. The current heart rate is shown in

Figure 3b. When the user starts a measurement, data

are continuously stored in the device and the related

heart rate chart is continuously plotted as shown in

Figure 3c.

5.5 Integration with EEGBase

The collected data can be transferred to any storage

system supporting the odML format. EEGBase (con-

trary to the original purpose EEGBase is newly con-

sidered as a storage and management system for any

kind of electrophysiology data), extended with an ad-

vanced system of user templates for storing various

kinds of data and metadata in the odML format, was

selected as a storage. Any template can be deﬁned ac-

MoBio - A Mobile Application for Collecting Data from Sensors

119

(a) List of available sensors (b) Current heart rate (c) Long term heart rate

Figure 3: Mobio Application Preview.

Figure 4: Data stored in EEGBase.

cording to data coming from a sensor and enriched by

a relevant set of metadata.

Technically EEGBase provides a RESTful API

containing methods for uploading newly collected

data. The user can select an automatic data upload

triggered when the user gets on-line. The user can

also enforce the upload manually. Once the data are

transferred to the server, the odML document is stored

in the ElasticSearch noSQL database and visualized.

Figure 4 shows visualization of the data from List-

ing 1.

6 DISCUSSION

The presented terminology covers a basic set of data

produced by typical sensors. Moreover, the terminol-

ogy is easily extensible since the odML open format

is used. Its JSON serialization is suitable for wireless

transfer due to low memory requirements in compar-

ison with XML. If we suppose a larger network of

sensors for human tracking (e.g. 15 sensors, each

sensor produces 10 signals with sampling frequency

50Hz, the sample size is 1B), then one hour record-

ing produces approximately 27MB of data plus a few

kilobytes for JSON elements. This data size is reason-

able with regard to the current capacity of SD cards.

The data security is ensured by the used protocols be-

cause both Bluetooth and ANT support data transfer

via a secured channel.

7 CONCLUSIONS

In this paper we observed a diverse collection of avail-

able sensors on the market. These sensors vary e.g. in

the used data format and data transfer protocol. We

identiﬁed architectures reﬂecting the ways the data

collected from sensors are managed. Then we se-

lected the three-layer sensor architecture as the most

suitable one for the long-term storage and process-

ing of sensory data. Since sensors producers use their

own systems and proprietary data formats, there are

difﬁculties with reading, transferring, and processing

data. As a solution we presented a prototype of a

custom terminology describing data collected from

sensors. When creating this terminology we used

ICT4AWE 2016 - 2nd International Conference on Information and Communication Technologies for Ageing Well and e-Health

120

and applied our expertise in neuroinformatics domain.

The terminology is implemented in the Android mo-

bile system MoBio that serves as a proof-of-concept

implementation supporting a limited set of sensors.

Then a complete process from collecting data by the

mobile system, storage of data in a universal format

respecting presented terminology, and ﬁnal storage of

data in EEGBase is outlined.

Our future work includes extension of the pre-

sented terminology to cover a broader collection of

health sensors. This includes also surface electrodes

capturing electrical signals from the human brain.

Then the extended mobile application will convert

the data transferred using Bluetooth or ANT+ proﬁles

to an odML document that satisﬁes this terminology.

The described data can be fully managed in the sys-

tems such as EEGBase. A new version of EEGBase

with a larger support of the presented terminology and

description of the complete terminology itself will be

released. This terminology may well serve to devel-

opers of similar systems when they want to collect

and process data from health sensors.

ACKNOWLEDGEMENTS

This publication was supported by the project

LO1506 of the Czech Ministry of Education, Youth

and Sports

REFERENCES

Davison, A. P., Brizzi, T., Guarino, D., Manette, O. F.,

Monier, C., Sadoc, G., and Fr

egnac, Y. (2013).

Helmholtz: a customizable framework for neurophys-

iology data management. Frontiers in Neuroinformat-

ics, (25).

eMarketer (2015). 2 billion consumers worldwide to get

smart(phones) by 2016. http://www.emarketer.com/

Article/2-Billion-Consumers-Worldwide-

Smartphones-by-2016/1011694.

Grewe, J., Wachtler, T., and Benda, J. (2011). A bottom-up

approach to data annotation in neurophysiology. Fron-

tiers in Neuroinformatics, 5(16).

HDF5 Group (2013). Hierarchical data format. http://

www.hdfgroup.org/HDF5/.

INCF (2013). International neuroinformatics coordinating

facility. http://www.incf.org/.

Jezek, P. and Moucek, R. (2012). System for EEG/ERP

Data and Metadata Storage and Management. Neural

Network World, 22(3):277–290.

Jezek, P. and Moucek, R. (2013). Eeg/erp portal for android

platform. Frontiers in Neuroinformatics, (46).

Le Franc, Y., Bandrowski, A., Bruha, P., Pape

z, V., Grewe,

J., Moucek, R., Tripathy, S. J., and Wachtler, T.

(2014a). Describing neurophysiology data and meta-

data with oen, the ontology for experimental neuro-

physiology. Frontiers in Neuroinformatics, (44).

Le Franc, Y., Gonzalez, D., Mylyanyk, I., Grewe, J., Jezek,

P., Moucek, R., and Wachtler, T. (2014b). Mo-

bile metadata: bringing neuroinformatics tools to the

bench. Frontiers in Neuroinformatics, (53).

Lowe, S. and Laighin, G. (2012). The age of the virtual

trainer. Procedia Engineering, 34:242 – 247.

Moucek, R., Bruha, P., Jezek, P., Mautner, P., Novotny, J.,

Papez, V., Prokop, T., Rond

ık, T.,

Stebet

ak, J., and

Vareka, L. (2014). Software and hardware infrastruc-

ture for research in electrophysiology. Frontiers in

Neuroinformatics, 8(20).

Seeger, C., Buchmann, A., and Van Laerhoven, K. (2011).

Wireless sensor networks in the wild: Three practical

issues after a middleware deployment. In Proceed-

ings of the 6th International Workshop on Middleware

Tools, Services and Run-time Support for Networked

Embedded Systems, MidSens ’11, pages 1:1–1:6, New

York, NY, USA. ACM.

Stoewer, A., Kellner, C., Benda, J., Wachtler, T., and Grewe,

J. (2014). File format and library for neuroscience

data and metadata. Front. Neuroinform. Conference

Abstract: Neuroinformatics 2014.

Teeters, J., Harris, K., Millman, K., Olshausen, B., and

Sommer, F. (2008). Data sharing for computational

neuroscience. Neuroinformatics, 6(1):47–55.

Teeters, J. L., Benda, J., Davison, A. P., Eglen, S., Ger-

hard, S., Gerkin, R. C., Grewe, J., Harris, K., Jackson,

T., Moucek, R., Pr

opper, R., Sessions, H. L., Smith,

L. S., Sobolev, A., Sommer, F. T., Stoewer, A., and

Wachtler, T. (2013). Considerations for developing

a standard for storing electrophysiology data in hdf5.

Frontiers in Neuroinformatics, (69).

Zehl, L., Denker, M., Stoewer, A., Jaillet, F., Brochier, T.,

Riehle, A., Wachtler, T., and Gr

un, S. (2014). Han-

dling complex metadata in neurophysiological exper-

iments. Frontiers in Neuroinformatics, (29).

MoBio - A Mobile Application for Collecting Data from Sensors

121