Advanced Multi-parametric Monitoring and Analysis for

Diagnosis and Optimal Management of Epilepsy and

Related Brain Disorders: The ARMOR Project

Christos P. Antonopoulos

, Nikolaos S. Voros

, André Bideaux

Panagiota Anastasopoulou

, Stefan Hey

and Wilhelm Stork

Technological Educational Institute of Western Greece,

Computer and Informatics Engineering Department,

Embedded System Design and Applications Lab,

National Road Antiriou-Ioanninon, 30020, Antirio, Greece, 30020, Greece

{chantonopoulos, voros}@teimes.gr

Karlsruhe Institute of Technology,

Institute for Information Processing Technologies, Engesserstr. 5, 76131 Karlsruhe, Germany

{panagiota.anastasopoulou, andre.bideaux, stefan.hey,

wilhelm.stork}@kit.edu

Abstract. The ARMOR project addresses the needs of the epileptic patient and

healthcare professional, aiming at the design and development of a non-

intrusive Personal Health System (PHS) for the monitoring and analysis of epi-

lepsy-relevant multi-parametric data, (i.e. EEG, EOG, EMG, EKG, skin con-

ductance data) and the documentation of the epilepsy related symptoms.

ARMOR platform incorporates models derived from data analysis based on al-

ready existing state-of-the-art communication platform solutions emphasizing

on security issues and required adaptations to meet ARMOR specifications. In

this context, this chapter aims to provide an extensive description of the main

aspects and issues addressed in the project as well as the main characteristics of

the developed platform.

1 Introduction

In this chapter we are introducing the main concepts of ARMOR EU funded project.

The main goal of the specific project is to manage and analyze a large number of

already acquired and new multimodal and advanced technology data from brain and

body activities of patients with epileptic disorders and controls (Magnetoencephalog-

raphy (MEG), multichannel Electroencephalography (EEG), video, Electrocardio-

gram (ECG), Galvanic skin response (GSR), Electromyography (EMG), etc) aiming

to design a more holistic, personalized, medically efficient and economical monitor-

ing system.

The ARMOR project effectively tackles requirements posed by both patients and

professionals, regarding a low cost yet highly efficient and secure ambulatory epilep-

sy monitoring platform. The platform is able to acquire as well as analyze (either

Hey S., Anastasopoulou P., Bideaux A., Stork W., Voros N. and Antonopoulos C.

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The ARMOR Project.

DOI: 10.5220/0006144500340055

In European IST Projects - The Quest for Excellence Towards 2020 (EPS Vienna 2014), pages 34-55

ISBN: 978-989-758-101-4

online or offline) epilepsy-relevant multi-parametric data, (i.e. EEG, EOG, EMG,

EKG, skin conductance data).

Epilepsy is a common, devastating and still incurable disorder. Although in most

cases its symptoms can be ameliorated by life-long pharmaceutical treatment, still this

treatment needs continuous adjustment and change to retain its efficacy. Due to its

multifactorial causes and paroxysmal nature, epilepsy needs multi-parametric moni-

toring for purposes of accurate diagnosis, prediction, alerting and prevention, treat-

ment follow-up and presurgical evaluation. The incidence of epilepsy is age-related,

higher in children; epileptic seizures occur in 1-2% of the general population and in

4% of children. During the periods of childhood and adolescence non-epileptic par-

oxysmal events (NEPE) also occur more frequently than in adult life with similar

clinical features. It is important to note that 30% of people with epilepsies have also

NEPE. Furthermore, epileptic seizures differ with respect to motor, cognitive, affec-

tive and autonomic and EEG manifestations. Their recognition and full understanding

is the basis for the optimal management (including additional diagnostic tests and

genetics) and treatment. The total cost of epilepsy in EU is counted upwards of 15

billion euros per year, with the severe impact on the patient of the social stigma and

the feeling of unpredictably seized, being unaccountable [1-3].

Current diagnostic methodologies and the need for advancement in this area com-

prise yet another important factor making epilepsy a prominent disorder to tackle.

Such methodologies include video EEG that records the habitual suspected event or

ambulatory EEG without video (for long term home recordings). Therefore, there is a

need for more accurate diagnosis of integrated seizure phenotype in individual pa-

tients, which will allow better understanding of underlying mechanisms, prediction

(and alert) of time and type of seizure (and alert) and availability of medical assis-

tance and advice [4-9].

In order to tackle the aforementioned challenges, ARMOR project developed an

ambulatory monitoring system for diagnosis and management, limited scalp EEG

covering and custom-designed multi-polygraphy (textile based EMG, body activity

sensors, autonomic and other biological data such as blood pressure, temperature,

sugar blood levels and O

and CO

saturation continuous monitoring). Diagnosis of a

disease as multifactorial and unpredictable as epilepsy demands continuous observa-

tion and correlation analysis of as many parameters as possible of the patient’s brain,

body and the environment.

In the context of ARMOR project, the above major medical problem has been ad-

dressed by employing the current advanced ICT technology and further advancement

in data analysis included in a way, which will benefit both the patient and the econo-

my of the health care system. In that respect, exploiting state-of-art wireless sensor

networking technologies ARMOR is envisioned as an ambulatory monitoring system

for diagnosis and management with video, limited scalp EEG covering and custom-

designed multi-polygraphy (textile based EMG, body activity sensors, autonomic and

other biological data such as blood pressure, temperature, sugar blood levels and O2

and CO2 saturation continuous monitoring). Therefore, issues such as efficient and

robust communication performance, minimization of power consumption and integra-

tion of different diverse technologies comprise cornerstones of such endeavor. Fur-

thermore, in the context of a complete end-to-end Personal Health System (PHS)

ARMOR platform includes the development of information and Tele-alarm Server as

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

well as client software for the ARMOR information and tele-alarm services. At the

same time emphasis is paid on secure communication between client software, home

gateway and wearable sensors while providing a wide range of health reporting ser-

vices and applications (Mobile, web-based and in-the-cloud back-office). From the

very beginning to the end of the project the theoretical background has also been

elaborated and the effectiveness of the ARMOR sensors has been improved by tar-

geted research work that proceeded in parallel with the steps described above. This

research involved sophisticated analysis of existing data from expensive devices (that

are not routinely available in clinics, e.g. multichannel EEG and/or MEG) and analy-

sis of selected data obtained during ARMOR project, which represent prototypical

examples or critical cases for diagnosis and classification.

2 System Architecture

Aiming towards a holistic, medically efficient and economical monitoring system

ARMOR platform addresses all functional aspects required. Thus, following the

overall system architecture as depicted in Figure 1 a critical segment comprises of the

sensors enabling the data acquisition in the local site (e.g. the patient's home envi-

ronment). Based on the extensive experience of ARMOR consortium as well as on

the equipment provided by the involved partners, multi-parametric data acquisition is

offered through a wide range of possible sensors gathering a wide range of medical

data continuously and in real-time.

Fig. 1. ARMOR Platform Architecture.

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

The first level of data aggregation and processing is done at the home gateway

where the ARMOR MiddleWare (AMW) provides access towards the upper modules

and vice versa. The role of the AMW is crucial since it comprises of the gateway

point between the sensor hardware equipment and the functional modules of software

application. Specifically, on one hand, support is provided for all types of data and

sensors utilized in the context of ARMOR, and on the other hand a wide range of

services is provided in order to support the depicted functional modules. Furthermore,

it functions as the communication bus among all required modules.

The ARMOR information server, as part of a complete remote Electronic Health

Record (EHR) system, is also a critical component of ARMOR system since it hosts

the models derived from extended research effort for multi-parametric data analysis.

However, such functionality requires close collaboration with patient data stored in

secure databases, which comprise another important subsystem of the ARMOR plat-

form. Another functionally critical module depicted is the ARMOR Application

Server. It provides to specialized personnel, like patients, medical stuff and caregiv-

ers, access to ARMOR system through a wide range of user interfaces. Such interfac-

es include visualization of multi-parametric data processing results, EHR access and

personal tracking or nutrition habits and vital signs information.

Finally, ARMOR emphasizes on security issues of sensitive medical data through

a specialized security layer where all ARMOR sub-systems area actively involved.

The first layer tackling security issues are sensors ensuring secure data acquisition

and transferring towards the aggregation point/s. All storage sites also employ securi-

ty techniques ensuring data integrity and privacy. Data communication and data trans-

ferring are challenging issues that have to be addressed, especially when they are

performed over the air. Both sensor communication and backhaul communication

parts of systems like ARMOR are susceptible to a wide range of dangers and possible

attacks requiring special attention. Finally, as far as the offline data processing and

data management center are concerned, access rights, user authentication and authori-

zation have also been taken into account as part of ARMOR EHR system.

2.1 Mobile Sensors for Multiparametric Monitoring in Patients with Epileptic

Disorders

As ARMOR’s main target is patients with epileptic disorders, it is vital to have EEG

sensors present in the system, as it is an essential component in the evaluation of

epilepsy. It has been shown that ambulatory long-term EEG recordings with intensive

monitoring have led to better classifications of seizures and treatment results [10].

Because 30-60% of all patients are unaware of their seizures, multiparametric moni-

toring can lead to new results with optimal treatment. Without EEG recordings, false

diagnosis may be made, as various phenomena are similar to the resulting behavior of

a seizure.

Electrocardiogram (ECG) is used to record the electrical activity of the heart. It is

an effective means to help to rule out a seizure being caused by the way the heart is

working. It has been noted that in some seizures, especially those located in the tem-

poral lobe, experience a change in heart rate prior to or at the onset of the seizure

[11]. A study showed an increase in heart rate of at least 10 beats per minute in 73 %

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

of seizures (93 % of patients) and this occurred most often around seizure onset. In

23 % of seizures (49 % of patients) the rate increase preceded both the electrographic

and the clinical onset [12]. Such changes may clarify the timing of seizure onset and

can be useful for seizure diagnosis and for automatic seizure detection.

Therefore the goal is to develop a multiparametric monitoring system that assists

in diagnosis, prognosis and treatment of the disease. Such system should fulfill the

following criteria: it should be non-invasive, mobile, continuous and unobtrusive and

all possible security and privacy aspects should be taken into account. This section

describes the design of the sensors for mobile epilepsy monitoring.

2.1.1 Sensor Requirements for Multiparametric Monitoring of Patients with Epileptic

Disorders

2.1.1.1 Functional Requirements

The functional requirement of ARMOR system include:

 Sampling Rate: The sampling rate for the physiological signals varies depending

on the physiological parameter that is assessed and based on the state of the art re-

search.

 Resolution: The output of the sensors offers at least 12-bit resolution.

 Data Format: The data format of the physiological sensor data that is used is

based on the unisens-format. This is a universal and generic format suitable for

recording and archiving sensor data from various recording systems and with var-

ious sampling frequencies.

 Interfaces: The sensors communicate with the aggregator wirelessly by using the

Bluetooth interface in order to transmit the pre-processed data for the online

analysis or/and via USB in order to store the assessed raw data.

 Online Analysis: The online analysis can be partially performed on the sensor

side, where the data are pre-processed and on the aggregator where a novel

Data

Stream Management System

(DSMS) effectively performs the further analysis that

needs extra computation power.

 Security: The communication between the sensors platform and the aggregator

employs encryption algorithms such as AES.

2.1.1.2 Non-functional Requirements

The non functional requirement of ARMOR system include:

 Weight, Dimensions, Housing: In order to develop a system appropriate for use in

everyday life, the sensor platform have to be as unobtrusive as possible. Of course

some technical limitations such as the battery consumption/ dimension might lead

to bigger housings. A trade-off between the comfort and the system lifetime has to

be made.

 Usability: Usability is a very important aspect that is also taken into account, as

the end-users (Doctor, Patient, Healthcare Professional, Family Member or Care-

giver) might be persons with limited technical knowledge. Therefore, the software

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

for starting/stopping the sensors offers an adequate GUI and special attention to

user friendliness is paid.

 Calibration & Service: Calibration and service must also be supported by the

producer/distributor of the specific unit.

 Distribution to the End-users: The patient will be wired-up at the hospital before

he/she goes off to spend his/her day. The electrodes will be firmly fixed by the

caregivers, whereas the sensors mostly needed for monitoring during nocturnal

sleep will be easily placed / replaced by the patients before he/she goes to bed and

unplugged in the morning after awakening.

 Time of Use: The sensors should be able to measure for at least 24 hours.

 Price: The price of the final system will depend on the number of units that will

have to be used.

2.1.2 The ARMOR Sensor System for Mobile Epilepsy Monitoring

In order to achieve the main requirement, the multi-parametric monitoring, different

sensors are integrated in the system. This includes not only the most relevant bio-

signal for epilepsy monitoring, the EEG but also ECG, GSR and a push button that

have be proven to be important in mobile epilepsy monitoring [13-15]. For this sys-

tem the following sensors have been selected.

The EEG module (Trackit

TM, Lifelines Ltd, Over Wallop, UK) is a mobile ambu-

latory device that can measure up to 32 channels with a sampling rate of 256 Hz.

Each channel has an ADC resolution of 16 Bit and has a maximum differential AC

input 10 mV. Depending on the number of channels, the sampling rate and the battery

used the device can achieve up to 96 hours of recording.

The ECG module (ekgMove, movisens GmbH, Karlsruhe, Germany) is a single

channel ECG recorder with a 12-bit resolution and a sampling rate from 256 Hz to

1024 Hz. The module can either be used with electrodes integrated into a wearable

chest strap, which is light, small and comfortable or be used with conventional dis-

posable wet electrodes. The electrodes of the chest strap are dry, allowing the every-

day use. To assess the patient’s physical activity, the module has also an integrated

triaxial acceleration sensor (adxl345, Analog Devices Inc.) with a range of ±8 g and a

sampling frequency of 64 Hz and an air pressure sensor (BMP085, Bosch GmbH)

with a sampling frequency of 8 Hz and a resolution of 0.03 hPa.

Fig. 2. ECG-Sensor with activity monitoring module.

The GSR module (edaMove, movisens GmbH, Karlsruhe, Germany) measures the skin

conductance with a sampling rate of 32Hz. The measurement range of the GSR module is

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

2μS to 100μS and its resolution is 14 bit. With this one can measure both the electrodermal

activity level (EDL) and responses (EDR). Furthermore the module has the same sensor

set to measure physical activity integrated as the ECG module.

Fig. 3. GSR Sensor.

The push button module (bioPLUX, PluX, Arruda dos Vinhos, Portugal) is a device

that collects up to 8 signals from various sensors and transmits the signals via Bluetooth to

a computer, where they can be viewed in real time. The resolution of bioPLUX is 12 bits,

and its sampling frequency is up to 1 kHz. For this system only one channel with a push

button sensor is used.

To assess additional context parameters like sound, light or geographic position, a

smartphone can be used. Connection between Smartphone, Sensor and Aggregator is

also realized by Bluetooth interface.

2.2 Communication Infrastructure

From communication point of view, ARMOR project developed a secure and effi-

cient platform that is able to acquire data, aggregate them to a gateway on the home

environment, exploiting state-of-the-art Wireless Sensor Network technologies and

finally convey them to a data repository residing into a hospital or a caregiver facility.

Consequently, a complete communication infrastructure comprises a multifaceted

objective in order to achieve efficient and secure end-to-end data handling and trans-

ferring.

Respectively, the whole process can be segmented into three major parts: (a) the

sensors data acquisition functionality, (b) the wireless transmission of data from the

sensor to the aggregation point, and finally (c) EHR for controlling data management

and data extraction from home gateway to the hospital facility.

2.2.1 Secure Sensors' Data Acquisition

Focusing on the sensors side, one key requirement is related to security. This is be-

cause respective devices comprise a potentially weak and vulnerable point in the

whole chain of data flow. Specifically, typical WSN sensors employed in the

ARMOR platform are small, low cost, devices operating unattended (e.g. in a pa-

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

tient's home environment) for extended periods of time thus being relatively easily

acquired by unauthorized persons trying to access sensitive data. ARMOR sensors

dispose of a Bluetooth interface, a debugger interface, a power supply and a local

storage on a SD card. The debugger is used to read or flash the sensor and to debug

the program. With respect to these main characteristics various alternatives have been

studied and adopted to enhance security provision. To avoid the read of the data or

the program code from the microcontroller, the debugger interface is disabled. The

sensor platform disposes of a special interface to update the flash with a program

code: the so-called “bootloader”. This interface gives no access to the storage and can

only be used to transfer a program. In the case of using an encryption of the local

storage it is advisable that this interface is disabled as well. Additionally, attacks over

the power supply are normally made in laboratory to get a secret key that is not

changing and need to use measurement instruments direct on the sensor. In the case

of a platform using data encryption, the secret key is changing periodically with re-

spect to new measurements. The secret key is needed while saving the data on the

local storage and so the user of the sensor will notice this kind of attacks.

ARMOR sensors are based on the MSP 430 microcontroller unit from Texas In-

struments offering a performance optimized software implementation of the Ad-

vanced Encryption Standard (AES). This implementation is designed for the 16-bit

RISC architecture of the Texas Instruments MSP 430 controller family and it is pro-

vide as a C interface offering high level security AES crypto features [16].

Research on similar commercial products has shown that no encryption mecha-

nism on the local storage is used [17,18]. Based on the above risk analysis encryption

on the local storage is not required.

2.2.2 Wireless Sensor Network Data Transmission

In the context of a demanding medical application, such as epilepsy monitoring, a

wide range of different signals leading to diverse data traffic requirements are re-

quired to be transferred. For example, in ARMOR project the sampling frequency of

signals acquired could span from few Hz or tens of Hz (e.g. accelerometer and respi-

ration sensor) up to hundreds of Hz or around 1 KHz (e.g. EEG and ECG signals).

Taking into consideration that each sample is typically represented by a 16-bit num-

ber as well as that in many cases multiple sensors are required in each case (e.g. a

complete EEG monitoring may require tens of EEG electrodes) it can be easily de-

duced that a wide range of traffic rates must be supported.

However, selecting an appropriate technology is not straightforward due to the

specific requirements posed by typical WSN sensors and platforms. During the last

few years, state of the art WSN technologies exhibit impressive advances advocating

respective system as a prominent solution for demanding applications. On one hand,

advances in hardware design, integrated embedded systems and miniaturization have

proven that extremely small devices, yet sufficiently powerful, can be implemented.

Additionally, such devices offer the capabilities to acquire data from the physical

world, store and process them, but even more importantly to transmit them wirelessly

using embedded transceivers. On the other hand, software developments have intro-

duced a completely new communication paradigm attracting high interest both from

academia and industry offering flexibility, rapid network deployment, self-

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

organization, distributed operation, power aware functionalities etc. All these capabil-

ities offered have led to significant advancements in various networking areas espe-

cially focusing on physical, MAC and routing layers of the ISO/OSI networking

model [17, 18] characterizing respective technologies.

However, offering the aforementioned characteristics comes at the cost of extreme

limitations in critical aspects, which comprise Achilles’ hill especially regarding pro-

cessing power, storage capabilities, communication bandwidth and most importantly

energy availability. The implementation of a typical WSN node is based on compo-

nents with extremely limited resources such as low processing capabilities (usually

provided by 16-Bit based Micro Controller Units), limited available memory (in the

area of 10 Kbyte RAM) and scarce energy availability; most prominent WSN plat-

forms base their operation on few AA batteries or even batteries with significantly

less capacity like Shimmer [19] platform that uses 450 mAh batteries. At the same

time, nodes are expected to operate unattended for extended periods of time.

Furthermore, in order to live up to the expectations of sensitive applications such

as the medical ones, another equally critical aspect that short range wireless commu-

nication technologies have to address is the provision of adequate security. It is com-

monly accepted that in sensitive nowadays applications it is of paramount importance

to guarantee that data gathered, stored and transferred comply with strict ethical and

legislative regulations as well as data privacy, data integrity and authentication of

communication parties [20]. In order to provide these features, we must rely on the

efficient execution of robust and state of the art cipher algorithms. In the context of

the ARMOR platform, Bluetooth protocol was employed for ARMOR WSN infra-

structure, while IEEE 802.15.4 was also considered and evaluated.

Bluetooth is a wireless radio specification designed to replace cables as the medi-

um for data and voice signals between electronic devices. The specification is defined

by the Bluetooth Special Interest Group (SIG), which is made up of over 1000 elec-

tronics manufacturers. Primarily intended for mobile devices, Bluetooth’s design

places high priority on small size, low power consumption and low costs. Bluetooth

specification seeks to simplify communication between electronic devices by auto-

mating the connection process.

Bluetooth radios operate in the unlicensed 2.4 GHz Industrial, Scientific, and

Medical application (ISM) frequency range. This frequency is already widely used by

all kind of devices such as microwave ovens, baby monitors, cordless telephones, and

802.11b/g wireless networking devices. In order to avoid interference from these

devices, Bluetooth uses a technology called spread spectrum frequency hopping.

Spread spectrum frequency hopping changes the transmission frequency up to 1600

times per second across 79 different frequencies. As a result, interference on any of

those frequencies will only last a fraction of a second. This, coupled with the limited

range of Bluetooth radio transmitters, results in a robust signal that is highly tolerant

of other devices sharing the same frequency.

Contrary to IEEE 802.15.4 based solutions, where all relative platforms are char-

acterized by analogous capabilities, Bluetooth based solutions vary significantly de-

pending both on the version of the protocol supported and even more on the specific

implementation's characteristics. Therefore, concerning data rates solutions covering

a wide range from 300 Kbps up to 1.5 Mbps can be found [19], [21], [22].

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

Furthermore, contrary to IEEE 802.15.4, which effectively leaves security support

to the higher ISO/OSI layer, Bluetooth offers a complete security specification with

respective advantages and disadvantages. Bluetooth security is based on three critical

services: authentication, authorization, and encryption. The authentication service is

supported by ensuring that a device seeking a connection is indeed the one it claims

to be. Authorization is the process that determines whether or not a requesting device

is allowed access to specific information or services. Encryption helps to ensure con-

fidentiality by protecting private data from being viewed by unintended recipients.

Bluetooth v2 devices, upon which ARMOR platform is based, can be set in one of

three different security modes. In security mode 1, no security measures are utilized.

Any other Bluetooth device can access the data and services of a device in security

mode 1. Security mode 2 enacts security measures based on authorization. In this

mode, different trust levels can be defined for each of the services offered by the

device. Security mode 3 requires both authentication and encryption. The security

features of the Bluetooth specification provide for secure communication at the link

level that comprises a weak point. Moreover, there are some weaknesses that need to

be considered. These weaknesses arise from the specification’s heavy reliance on

device authentication for security services as well as the level of control that the user

has over Bluetooth devices and their configuration. The current Bluetooth specifica-

tion does not provide any means of user authentication. The lack of any means of user

authentication coupled with the reliance on device authentication leaves Bluetooth

particularly vulnerable to spoofing attacks and the misuse of authenticated devices

[23].

Based on projections indicating the rapid increase of short range wireless commu-

nication devices (e.g. in the context of a cyber physical system) the need for more

secure systems using stronger cipher algorithms, while not requiring excessive re-

sources, will also increase. Even more, provision should be made for cases where

new communication technologies will appear or when existing technologies (e.g.

Bluetooth) will need to coexist with new ones without compromising security level.

In such cases, an implementation not being limited by cipher algorithms of particular

communication protocol (e.g. E0 and E1 of Bluetooth) but rather following a more

widely adopted paradigm (e.g. AES) will be of significant added value.

In order to deal with the aforementioned challenges, in the context of ARMOR an

ultra-low power hardware encryption module has been developed. It provides high

security level regardless of the underlying communication technology employed and

is based AES encryption algorithm [49]. AES algorithm has been standardized by the

National Institute of Standards and Technology (NIST) as a highly secure block ci-

phering method. It has replaced the old DES algorithm, whose key sizes were becom-

ing too small. The developed encryption module implements AES algorithm in FPGA

technology and is highly optimized for ultra-low power dissipation rendering it an

ideal solution for WSN network applications. The encryption module has been care-

fully designed to require minimum logic resources as well as utilizing power aware

design techniques in architectural (8-bit datapath, use of sequential structures, re-

source reuse, optimized Galois Field Multiplier which is the structural datapath ele-

ment of the cryptographic engine, pipelining, path balancing, one hot FSM encoding)

as well as in FPGA implementation level (clock and data disabling). The use of these

techniques has resulted in a significant performance/silicon footprint ratio [24].

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

2.2.3 Personal/Electronic Health Record Technologies

In the context of a holistic approach like the one offered by ARMOR project, the

adoption of the proper EHR technology is of paramount importance. As part of

ARMOR system it entails all aspects of data handling, management and transferring

from the home environment of the user (i.e. the patient) to the service provider that

can be a hospital, a clinic, a doctor or any kind of care giving facility. ARMOR’s

Personal Health Records (PHR) intends to be a standard electronic management sys-

tem of medical information exchanged on between the patient and his/her physicians,

as well as amongst medical organizations that collaborate for providing integrated

medical care services. ARMOR PHR has been designed using the latest technology

that allows efficient and secure handling of patients’ sensitive medical data.

Significant effort has been devoted to security aspects since in today’s global and

mobile economy, it is common to hear stories of companies paying significant costs

and exposing themselves to lawsuits due to corporate assets and data being compro-

mised. At the same time, IT departments started to face new risks associated with the

growing need to support remote access to mobile users, partners and customers. Ad-

ministrators must constantly compromise the risks associated with providing remote

access with increasing demands for mobility. Remote users require access to compa-

ny’s resources any time from anywhere, using any computer whether it is provided by

a company, own laptop or an internet cafe PC, mobile or handheld PDA device. E-

Health systems, platforms and services are also examples of such requirements. They

employ standard networking technologies and hence they are vulnerable to same

types of attacks. Having to deal with extremely sensitive private and personal data,

they are also exposed at the risk of law suits and they need to employ much more

stringent security mechanisms, including data acquisition from distributed medical

instruments and protection of (often) distributed EHR repositories against intrusion

and data theft.

The end-to-end security is one of the possible approaches to counter fight these

risks. End-to-end security means that sensitive data are encrypted all the way from the

device side application back to the enterprise. Rather than relying on transport-level

security (such as Secure Sockets Layer, or SSL), end-to-end security puts the power

of strong encryption in the user's hands, all through a simple interface. The specific

requirements posed for the technology adopted in ARMOR included: authenticated

and authorized sensor information access, secure Web service interfaces to PHR and

secure Web services with HTTP.

Nowadays the area of EHR technologies has attracted high interest as indicated by

the wide range of standardization bodies involved [25]. The data structure algorithms

and the user iInterfaces proposed [26-28] as well as the number of different and di-

verse implementations [29-33]. The ARMOR platform has adopted the intLIFEPHR

solution provided by Intracom S. A. Telecom Solutions, which is member of the

ARMOR consortium. intLIFEPHR consists of the following subsystems:

 Electronic Health Record Subsystem

 Vital Signs Monitoring Subsystem

 Personal Health Record Subsystem

 intLIFE Management Subsystem

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

Through the above features intLIFEPHR perfectly suited the needs of the ARMOR

platform offering an efficient, robust and secure backhaul connection between the

home environment and caregivers' facility.

2.3 System Middleware

The ARMOR Middleware (AMW) represents the ICT component that provides the

necessary infrastructure to acquire & store locally and upload to remote server (the

PHR) sensor data. It provides infrastructure for real time computation of raw sensor

data (modalities) along with the necessary data aggregation and synchronization func-

tions. It also provides a notification module that communicates specific events (main-

ly alarms) to high-level applications

AMW consists mainly of three parts. The first part is the xAffect that collects all the

data from the sensors and fuses them to a synchronized data stream. The second part is the

graphical user interface. It is used to interact with the patient or healthcare professional

and also handles the data storage and uploads. The last part is the Data Stream Manage-

ment System (DSMS), which it uses the data stream to perform online analysis to detect

events of special interest.

2.3.1 xAffect

xAffect is a software framework developed by the Research Center for Information Tech-

nology, Karlsruhe, Germany. It has been developed in Java to fulfill real-time data pro-

cessing, easy integration of different data sources, easy integration of algorithms and data

logging of raw as well as derived data [34]. The data format, which is being used, is the

unisens-format. This is a universal and generic format suitable for recording and archiving

sensor data from various recording systems and with various sampling frequencies [35].

Although xAffect™ offers much functionality, in the context of ARMOR two

main parts of it have been tailored to ARMOR needs: data streaming and data re-

cording, which fulfill AMW’s integration requirements successfully. However the

capabilities for data fusion remain available as complement to the data stream man-

agement system, named ARMOR Insight.

Version 1.01.846 of xAffect™ allows to use it as library, which resulted in em-

bedding it as a component in AMW. The specific version has also been modified in

order to customize the interface with AMW that was necessary to achieve the perfor-

mance and the functionality required for the AMW. The changes that were introduced

can be summarized as follows:

 Additional libraries for bioPlux and TrackIT. In order to use a broad spectrum of

sensors, non-existing libraries had to be written from scratch.

 A decryption module that allows realizing ciphered data coming from ARMOR

sensors will come ciphered.

 Data acquisition pause/resume to achieve the needs for the control of the sensor

data acquisition.

 A custom notification module for communicating xAffect state to AMW DSMS.

 Data recording functionalities have been extended to provide configurable file

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

splitting (in order to reduce high network consumption during heavy data up-

loads), alarm signal detection and a communication system with AMW PHR up

loader daemon.

 Data streaming functionalities have been extended to provide hot-plug client con-

nections and custom xml output data formats (including gzip for network traffic

optimization).

2.3.2 Graphical User Interface

In the Graphical User Interface (GUI) the user enters his/her username and password

and the patient-id, which corresponds to a personalized profile from the Server. Af-

terwards the user can press the configure button to initialize the communication with

the sensors. Furthermore the profile contains information about the alarm settings.

This enables the system Middleware to set up the DSMS with the customized set of

alarms [36].

Fig. 4. ARMOR Graphical User Interface.

When the configuration process has finished successfully the user is able to start

the measurement by pressing the record button. During recording, the data are

streamed from the sensors through xAffect towards the DSMS and the local storage.

The GUI also allows the user to pause or resume the measurement. This allows the

subject to interrupt the data acquisition and move out of the Bluetooth range of the

Home Gateway.

2.3.3 Data Stream Management System (DSMS)

The DSMS function takes place on-line, where real time processing of modalities is

performed. The DSMS is based on Microsoft™ StreamInsight™ platform created for

the development and deployment of complex event processing (CEP) applications.

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

It’s a high-throughput stream processing architecture that uses .NET Framework-

based development platform.

The development done at the DSMS allows receiving xAffect™ sensor data in re-

al time and forewards them in a lossless way to the computation algorithms. In this

way, the framework is able to create queries and policies over data.

Lossless data streaming between xAffect™ and DSMS is achieved by employing

TCP channel. This makes sure the built-in architecture of StreamInsight™ can work

with the synchronized data coming from xAffect™.

The main objective of the DSMS is to deliver alarm and warning events prede-

fined in the profile. One of the most important events is the push button detection.

Other events are for example an alpha rhythm or seizure detection.

3 Epilepsy Monitoring and Data Analysis

The proposed system provides novel functionality for both real-time (online) and

offline analysis of data. New techniques have been developed for multi-parametric

sensor data mining (data fusion and correlation analysis, integration of information

from various data sources/modalities, similarity analysis of signals, clustering, classi-

fication and prediction). Novel real time (online) analysis methods for multi-

parametric stream data have also been developed aiming at detecting signals beyond

the limits, identify seizure premonitory signs, discover typical patterns of activity

followed by seizures and detect any typical patterns of activity/behaviour based on

models that will be created. Trade-offs for automated analysis are taking place at the

local site of each patient (instead of at the Health Center) aiming to reduce processing

time, storage requirements and communication cost, facilitating the reduction of raw

data to secondary and tertiary parameters (that have been correlated), have also been

taken into account. All analysis and emergency alert mechanisms are based on a per-

sonalized model according to the patient's health profile. New decision support tools

for advising the patient, triggering an alarm and detecting emergency situations have

also been developed.

In addition, new informatics tools have been developed for offline analysis of

multi-parametric data correlation with other stored data about the patient (EEG, PET,

SPECT, fMRI, genetic data) and the disease, offline data fusion for certain combina-

tions of modalities (e.g. MEG, MRI) taking place at the Health Center with the partic-

ipation of medical experts as well as new functionality that will provide feedback to

the online analysis model. ARMOR has also contributed with novel contributions in

the analysis of multidimensional time series, similarity analysis of signals, detection

of patterns and associations between external indicators and mental states, analysis of

associations among signals and symptoms, discovery of lag correlation among differ-

ent signals, detection of vital signs of a person changing in a significant manner,

identification of motifs (in spatio-temporal signals) and frequently repeated patterns

or outliers (corresponding to seizure signs), and automatic summarization of results

for each patient.

Moreover, new techniques have been investigated for the detailed offline tomo-

graphic analysis of multichannel EEG and MEG data recorded simultaneously with

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

measurements of heart activity (EKG), Galvanic skin Response (GSR) and other

measurements that can be easily incorporated in the online monitoring for normal and

epileptic patients in awake state and at different sleep stages obtaining the most direct

insight of what is happening in the brain. As far as sensor data acquisition and pre-

processing (cleaning, integration, transformation and reduction) are concerned, both

existing and novel techniques for data reduction and summarization (to deal with data

streaming) have been considered.

Finally, in the context of ARMOR, existing database technologies have been ex-

tended to support the organization of multi-parametric data including the support of

efficient storage and retrieval capabilities such as multidimensional indexing. The

ARMOR databases store logs of all events, recorded values from sensors and other

metrics that are monitored, personalized patient health profiles, medical information

including guidelines for diseases, symptoms, medication, potential side effects of

medication, etc.

3.1 Current Practices and Challenges for Epilpsy Monitoring

Unprovoked seizures and epilepsy are fairly common treatable neurological condi-

tions (incidence of unprovoked seizures 33 to 98 per 100,000 per year; incidence of

epilepsy 23 to 190 per 100,000 per year; prevalence of epilepsy 3 to 41 per 1000;

lifetime risk of epilepsy: 1 to 3%) [37]. There is considerable disagreement about the

recurrence risk following a first seizure. Estimates of the recurrence rates following

the first seizure over two and three years have varied between 23% [38] and 71%

[39]; the risk of recurrence has been estimated at 14% at one year, 29% at three years

and 34% at five years [40]. In a systematic review and meta-analysis including both

prospective and retrospective observational studies, the pooled estimate of the risk of

recurrence of a first unprovoked seizure at two years was 42% (95% CI 39 to 44)

[41]. The more seizures an individual have had, the higher the risk of subsequent

seizures; the risk of a recurrence following two seizures is approximately 73% and

after three seizures is 76%. There is evidence that early treatment can reduce the risk

of seizure recurrence, and its efficacy depends largely on the appropriate drug choice

in relation to the particular clinical syndrome. Therefore, early and accurate diagnosis

of epilepsy is crucial in patients’ management. However, there are two important

clinical problems to consider.

First, the initial symptom of epilepsy (the first seizure in life) usually manifests as

an episode of loss or impairment of consciousness, usually associated with change of

muscle tone. The diagnostic challenge here is to distinguish an epileptic seizure from

other medical conditions that present with similar clinical features but require com-

pletely different and specific for each clinical category treatment and management.

These clinical entities manifest as non-traumatic transient loss of consciousness, i.e. a

brief clinical episode, characterised by rapid loss of normal responsiveness, loss or

reduction of muscle tone or stiffness and amnesia for the event [42].

Second, the generic term “epilepsy” is unsatisfactory for clinical use. Intense clin-

ical and genetic research over the last few decades have identified a large number of

well defined epilepsy syndromes with different clinical, EEG, neuropsychological,

and neuroimaging profiles, natural history and prognosis, conditions that ultimately

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

require different management. Identification of the particular form or epilepsy syn-

drome is the cornerstone of meaningful, optimal management.

Therefore, there are two major steps in the orderly diagnostic process of the pa-

tient with possible new onset epilepsy: 1) is it epilepsy or another disorder of Transi-

ent Loss of Consciousness (TLC)? and 2) what type of epilepsy is it?

Syncope is the commonest cause of TLC, due to cerebral hypoperfusion. Neurally

mediated (vasovagal, neurocardiogenic or reflex) syncope can happen in up to 40%

of the general population and can be misdiagnosed as epilepsy, particularly when it

results to cerebral hypoxia and to a reflex anoxic seizure. Cardiogenic syncope is due

to structural heart disease or arrhythmia, and psychogenic syncope can mimic organic

types of syncope [43].

Detailed history regarding possible triggers and the circumstances of the event and

accounts of several episodes from patients and onlookers are essential for the diagno-

sis of syncope and its differentiation from epilepsy and other causes of TLC. Physical

(heart auscultation and BP measurements) and neurological examination may reveal

specific cardiac or autonomic disorders and prompt the relevant referrals. An EEG is

not a basic ancillary test, unless a diagnosis of epilepsy is likely; however, on rare but

well documented occasions, focal epileptic seizures (mainly right temporal) may

trigger cardiac asystole and anoxic seizures. Important laboratory tests include:

 Electrocariogram (ECG) which determines the cause in less than 5% of cases [44].

 Echocardiography, of undetermined but generally small diagnostic yield (detects

structural cardiac abnormalities) [45].

 Exercise test, of less than 1% diagnostic yield. Prolonged monitoring for 48h

(Holter) or few weeks (continuous loop recorders) or months (invasive Medronic

device planted subcutaneously). A frequent problem is detection of arrhythmias

without symptoms.

 Tilt table test (positive in 50% of patients with syncope) [46].

 Other autonomic function test (of undetermined but generally small diagnostic

yield, time consuming and expensive).

Diagnosis of epilepsy may also require differentiation from other paroxysmal events

that may alter or appear to alter neurological function to produce motor signs or sen-

sory, autonomic or psychic symptoms that at least superficially resemble those occur-

ring during epileptic seizures. Such clinical events are typically known as non-

epileptic seizures (NES), and can be either of physiological (PhNES) or psychogenic

(PsNES) origin; distinction between these two types is important for proper treatment

and management and relies on recognition of organic symptoms and signs. More than

30% of the patients referred to epilepsy centers have NES only whereas a smaller

proportion has epileptic and NES in combination, particularly PsNES.

In order to deal with the aforementioned challenges, ARMOR allows progress be-

yond the above diagnostic and follow-up procedures by:

 Combining all appropriate measurements of brain and body activity at the same

time.

 Integrating these multi-parametric and multimodal data so as to allow better dif-

ferential diagnosis of epileptic from non-epileptic seizures as well as to define bet-

ter the type of epilepsy seizures represent.

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

 Allowing a custom made selection of the most simple and economical sensors to

be employed in each patient, based on (a) advanced analysis of data with initial

extended use of many sensors and (b) experience gained from advanced analysis

of similar cases.

 Allowing the final monitoring to be conducted at the patient's (child's) own physi-

cal environment and in a cheaper and more efficient way (for clinical purposes).

 ARMOR also provides a valuable clinical tool to clinical epileptologists as it

solves several different current technical problems of long term monitoring and

communication of the many parameters needed to describe the complex nature of

paroxysmal seizures.

 ARMOR through basic and clinical research has shed light to our ignorance on

several questions of epileptology which demand long term and accurate multi-

modal and multi-parametric monitoring including: (a) electroclinical correlations

of loss of consciousness in generalized seizures, (b) sleep/epilepsy relationships

and (c) stress and other triggers/premonitory signs of epileptic seizures.

 The possible relationship of above to the localization of epileptic foci activations

in brain space - with implications to pathophysiology of epilepsy and its therapeu-

tics including presurgical evaluation.

3.2 Offline Data Analysis Algorithms

The offline data analysis in ARMOR has two main goals: to help accurate diagnoses

and to support the online monitoring of patients. Various methods have been devel-

oped. These newly developed and other, already available, methods have been used to

analyze data from patients with epileptic disorders and healthy individuals, including

whole night polygraphic recording. The tools for polysomnography have offered

valuable experience in dealing with issues related to data from long term monitoring

(8 hours). Beyond the known strong interdependencies between sleep and epilepsy

[47], additional justification for studying normal sleep microstructure features has

been given by recent findings linking epileptic ictogenesis to sleep and more specifi-

cally to sleep instability [48], sleep K-complexes and sleep spindles. The results of

these offline data analyses provided new insights into different aspects of epilepsy

and sleep, and were relied upon to offer recommendation of the polygraphic online

monitoring of patients.

Every online recording performed with the ARMOR online platform system is up-

loaded to the Patient Health Record (PHR) database. From there it is transferred to

the offline analysis database. In order for this service to be precise, there are several

steps that should be followed. The first one is to check if there is a new recording in

PHR. It should be noted that every recording is related to a specific patient, a specific

device performing the recording and the data of the recording. The synchronization

service checks in specified time intervals for a new recording in the PHR system. If a

set of new recordings is detected, then these recordings are requested from the PHR

database. The format of the data in each new recording is Unisens. Once the whole

recording has been downloaded and checked for errors (e.g. download errors, missing

files etc) the data are transformed in EDF format. Once this step is completed a sub-

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

routine will handle the EDF data to extract all the necessary info and store them in the

offline analysis database. Then the newly uploaded data can be accessed through pre-

specified queries.

In the following figure the structure of the main components of the synchroniza-

tion of the two databases is presented.

Fig. 5. Main components of the database synchronization.

3.3 Online Data Analysis Algorithms

One way of performing an online analysis of sensor data is a multi-layer Data Mining

Model. These models are divided into four layers: data collection, data management,

event processing layer and data mining service layer. The ARMOR Online platform

is consisted of a similarly structured system. The data collection layer includes the

sensors and uses the xAffect tool for streaming the sensors’ data to a StreamInsight

application. The streamed data are being processed with a previously specified and

parameterised algorithm, before the detected events are extracted.

Fig. 6. Interface between different data management modules.

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

The core of each detection algorithm is implemented in Mathworks' Matlab envi-

ronment. Matlab offers many features that make the design of an algorithm more

efficient and effective as well as shortening the time needed for the development of

the algorithm. An online application on the other hand requires a Data Stream Man-

agement System (DSMS), which offers the ability to process online data in a time and

memory efficient way. The system used in our applications is StreamInsight from

Microsoft. To introduce our algorithms to the Microsoft's StreamInsight environment,

the .NET compiler and API of Mathworks have been used. In more detail, Mathworks

provides a compiler for .NET packages. The result of this procedure is a set of librar-

ies containing the algorithms and all the necessary components (functions, data such

as training models, etc) necessary for each algorithm. The compiled libraries can be

accessed by a StreamInsight application using the API that Mathworks provides. This

API is the Matlab Compiler Runtime, which provides all the necessary components in

order to use the algorithm in a stream application.

It should be noted that each algorithm has been designed to operate with segments

of the data, as it is necessary in an online application. In order to pre-process, the

streamed data, before having them processed from the detection algorithm, a set of

StreamInsight tools were used. Depending on the procedure followed by the detection

algorithm, the streamed data should be aggregated or transformed in matrices whose

rows and columns correspond to channels and samples of each segment, respectively.

The detection procedure is performed after this step. In the case of the seizure detec-

tor for each time window of the streamed data, a matrix should be formed. Each row

of this matrix contains the data values for one channel during the specific time win-

dow, whereas the columns of this matrix contain the values for all channels for a

specific time point. This transformation is possible due to StreamInsight's service of

User Defined Operator.

4 Conclusions

As detailed in the previous sections, ARMOR project has addressed various critical,

challenging and multifaceted objectives in order to effectively acquire, manage and

analyze large number of signals related to epilepsy monitoring and decision-making.

In order to offer such capabilities significant research, design and development efforts

have been devoted to different and diverse areas ranging from pure medical research

to engineering areas such as wireless sensor networking and data processing algo-

rithms.

In that respect, this chapter attempted to highlight the main issues tackled and pre-

sented the main achievements as well as the design aspects of ARMOR project. Ini-

tially, the main architectural goals of ARMOR platform were presented, focusing on

its main components. Then, the requirements and characteristics of the sensors used

was also presented, as well as the overall communication infrastructure employed for

robust, efficient and secure end-to-end transfer of user sensitive personal data. One of

the cornerstone components of ARMOR architecture, ARMOR middleware, was also

detailed. ARMOR middleware resides in the home gateway of ARMOR, plays multi-

ple roles in the monitoring process and provides diverse functionalities both related to

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

communication and data processing. The second part of the chapter focused on data

analysis processes and algorithms designed and developed in the context of the pro-

ject so as to gain significant insight to epilepsy monitoring and decision making.

Concluding, ARMOR project proposed a highly efficient system able to offer sig-

nificant assistance and advancements on epilepsy monitoring and decision-making

exploiting state-of-art technologies as well as extending them.

References

1. Christian M. Korff. Ingrid E. Scheffer, "Epilepsy classification: a cycle of evolution and

revolution", Curr Opin Neurol. 2013 Apr;26(2):163-7. doi: 10.1097/

WCO.0b013e32835ee58e.

2. ANNE T. BERG, J. HELEN CROSS, "Classification of epilepsies and seizures: historical

perspective and future directions", Handbook of Clinical Neurology, Vol. 107 (3rd series)

Epilepsy, Part I H. Stefan and W.H. Theodore, Editors, 2012 Elsevier B.V

3. Newton CR, Garcia HH, "Epilepsy in poor regions of the world", Lancet. 2012 Sep

29;380(9848):1193-201. doi: 10.1016/S0140-6736(12)61381-6.

4. Badea A, Kostopoulos GK, Ioannides AA. Surface visualization of electromagnetic brain

activity.J Neurosci Methods. 2003 Aug 15;127(2):137-47

5. Barnathan M, V. Megalooikonomou, C. Faloutsos, F. Mohamed, S. Faro, «High-order

Concept Discovery in Functional Brain Images», 7th IEEE International Symposium on

Biomedical Imaging: From Nano to Macro (ISBI), Rotterdam, The Netherlands, 2010, pp.

664-667.

6. Barnathan M, V. Megalooikonomou, C. Faloutsos, F.B. Mohamed, S. Faro, «TWave:

High-Order Analysis of Spatiotemporal Data», In Proceedings of the 14th Pacific-Asia

Conference on Knowledge Discovery and Data Mining (PAKDD), Hyderabad, India, June,

21-24, 2010, Advances in Knowledge Discovery and Data Mining, Lecture Notes in Com-

puter Science, 2010, Volume 6118/2010, pp. 246-253.

7. Iasemidis LD, Shiau DS, Pardalos PM, Chaovalitwongse W, Narayanan K, Prasad A,

Tsakalis K, Carney PR, and Sackellares JC. Long-Term prospective on-line real-time sei-

zure prediction. Clinical Neurophysiology 2005; 116(3):532-44.

8. Ioannides AA, Corsi-Cabrera M, Fenwick P, del Rio Portilla Y, Laskaris N, Khurshudyan

A, Theofilou D, Shibata T, Uchida S, Nakabayashi T and Kostopoulos GK.MEG tomogra-

phy of human cortex and brainstem activity in waking and REM sleep saccades. Cerebral

Cortex, 2004;14(1):56-72.

9. Velis D, Plouin P, Gotman J, da Silva FL; ILAE DMC Subcommittee on Neurophysiology.

Recommendations regarding the requirements and applications for long-term recordings in

epilepsy. Epilepsia 2007; 48(2), 379-84

10. Stefan H., Hopfengärtner R.: Epilepsy monitoring for therapy: Challenges and perspec-

tives. Clinical Neurophysiology 2009; 120:653-58.

11. Jones V.M., et al.: Biosignal and context monitoring: distributed multimedia applications of

body area networks in healthcare. 10th Workshop on Multimedia Signal Processing in

Cairns, 2008.

12. Zijlmans M., Flanagan D., Gotman J.: Heart rate changes and ECG abnormalities during

epileptic seizures: prevalence and definition of an objective clinical sign. Epilepsia 2002;

43: 847–54.

13. Massé F., Penders J., Serteyn A., Bussel M. van, Arends J.: Miniaturized Wireless ECG-

Monitor for Real-Time Detection of Epileptic Seizures. In: Wireless Health’10, San Diego,

USA (2010)

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project

14. Nijsen, T. M., Arends, J. B., Griep, P. A., Cluitmans, P. J.: The potential value of three-

dimensional accelerometry for detection of motor seizures in severe epilepsy. In: Epilepsy

& Behavior 7, 1, 74–84 (2005)

15. Poh M.-Z., Loddenkemper T., Swenson N., Goyal S., Madsen J., Picard R.: Continuous

Monitoring of Electrodermal Activity During Epileptic Seizures Using a Wearable Sensor.

In: 32nd Annual International Conference of the IEEE EMBS Buenos Aires, Argentina

(2010)

16. Texas Instrumments MSP430 microcontroller crypto software features infor-

mations:http://jce.iaik.tugraz.at/sic/Products/Crypto_Software_for_Microcontrollers/Texas

_Instruments_MSP430_Microcontrollers

17. Somnomedics: http://somnomedics.eu/home.html

18. Medset telesmart: http://www.medset.com/index.php?id=45

19. Shimmer platform http://www.shimmer-research.com/

20. EU Directive: Data Protection Directive (Directive 95/46/EC), Directive on Privacy and

Electronic Communications (2002/58/EC), Directive 2011/24/EU of the European Parlia-

ment and of the Council of 9 March 2011 on the application of patients’ rights in cross-

border healthcare

21. MoviSens: http://www.movisens.com/en

22. Revoing2011: Revoking Networks "Advanced User Manual" Version 4.77 2/3/2011

23. SANS Institute InfoSec Reading Room , "Brush up on Bluetooth", 08/21/2003

24. George Krikis, Christos P. Antonopoulos, Nikolaos S. Voros, "Design and Implementation

of Efficient Reconfigurable Cipher Algorithms for Wireless Sensor Networks" 11th IEEE

International Conference on Industrial Informatics, Bohum, Germany, 29-31, July, 2013

25. Eichelberg M, et al. (2005), “A survey and analysis of Electronic Healthcare Re-cord

standards”, ACM Computing Surveys. 37(4):277–315

26. CEN/TC251–ISO/TC215 (2010), I. E. Electronic Healthcare Record (EHR) Communica-

tion. Parts 1: Reference Model, Part 2: Archetype Model, Part 3: Reference Archetypes and

Term lists, Part 4: Security and Part 5: Inter-face Specification. 2010

27. Rector, A. et al. (1993), A framework for modelling the electronic medical record. s.l. :

Methods Information Medical, 32: 109-119

28. Beale, T. (2002), “Archetypes constraint-based domain models for future-proof in-

formation systems”, OOPSLA-2002 Workshop on Behavioural Semantics

29. Microsoft HealthVault (http://www.healthvault.com)

30. World Medical Card (www.wmc-card.com)

31. INDIVO Health (indivohealth.org)

32. TOLVEN Patient/Clinician HR (www.tolven.org)

33. OpenMRS (http://www.open-emr.org)

34. Schaaff K., Mueller L., Kirst M., http://www.xaffect.org/

35. Kirst M., Ottenbacher J., www.unisens.org

36. A. Bideaux, P. Anastasopoulou, S. Hey, A. Cañadas, A. Fernandez: Mobile monitoring of

epileptic patients using a reconfigurable cyberphysical system that handles multi-

parametric data acquisition and analysis: MobiHealth 2014

37. Hauser WA, Hesdorffer DC. Epilepsy: frequency, causes and consequences. New York:

Demos Publications, 1990.

38. Pearce JL,Mackintosh HT. Prospective study of convulsion in childhood. The New Zealand

Medical Journal 1979;89:1–3.

39. Elwes RCD, Chesterman P, Reynolds EH. Prognosis after a first untreated tonic-clonic

seizure. Lancet 1985;2:752–3.

40. Hauser WA, Rich SS, Annegers JF, et al. Seizure recurrence after a first unprovoked sei-

zure: an extended follow-up. Neurology 1990; 40:1163–70.

41. Berg AT, Shinnar S. The risk of seizure recurrence following a first unprovoked seizure: a

quantitative review. Neurology 1991;41:965–72.

EPS Vienna 2014 2014 - European IST Projects - The Quest for Excellence Towards 2020

42. Crompton DE, Berkovic SF. The borderland of epilepsy: clinical and molecular features of

phenomena that mimic epileptic seizures. Lancet Neurol 2009; 8: 370-81.

43. Stephenson JBP. Fits and faints. London: MacKeith Press, 1990; Lempert T. Recognizing

syncope: pitfalls and surprises. J R Soc Med 1996; 89: 372-5.)

44. Hammill SC. Value and limitations of non-invasive assessment of syncope Cardiol Clin

1997; 15: 195-218

45. Kapoor WN. Evaluation and outcome of patients with syncope Medicine 1990; 69: 160-75

46. Kapoor WN, Smith MA, Miller NL. Upright tilt testing in evaluating syncope: a compre-

hensive literature review. Am J Med 1994; 97: 78-88

47. Kokkinos V, Koupparis A, Stavrinou M and Kostopoulos GKThe hypnospectrogram: An

EEG power spectrum based means to concurrently overview the macroscopic and micro-

scopic architecture of human sleep. Journal of Neuroscience Methods 2009; 185(1):29-38.

48. Halasz P, Kelemen A. Relationship of phasic sleep phenomena, spike-wave discharges, and

state-dependent responsiveness in sleep. Epilepsia. 2010 May;51(5):934-5.

49. Eric Conrad, “Advanced Encryption Standard”, WhitePaper

Advanced Multi-parametric Monitoring and Analysis for Diagnosis and Optimal Management of Epilepsy and Related Brain Disorders: The

ARMOR Project