Conference PaperPDF Available

Abstract and Figures

The traditional hospital setup is not appropriate for long-term epilepsy seizure detection in naturalistic ambulatory settings. To explore the feasibility of seizure detection in such a setting, an in-hospital study was conducted to evaluate three wearable devices and a data collection platform for ambulatory seizure detection. The platform collects and processes data for study administrators, clinicians and data scientists, who use it to create models to detect seizures. For that purpose, all data collected from the wearable devices is additionally synchronized with the hospital EEG and video, with gold-standard seizure labels provided by trained clinicians. Data collected by wearable devices shows potential for seizure detection in out-of-hospital based and ambulatory settings.
Content may be subject to copyright.
Poster: RADAR-base: Epilepsy Case
Study
Zulqarnain Rashid
Institute of Psychiatry,
Psychology & Neuroscience
King’s College London
London, SE5 8AF, UK
zulqarnain.rashid@kcl.ac.uk
Yatharth Ranjan
Institute of Psychiatry,
Psychology & Neuroscience
King’s College London
London, SE5 8AF, UK
yatharth.ranjan@kcl.ac.uk
Callum L Stewart
Institute of Psychiatry,
Psychology & Neuroscience
King’s College London
London, SE5 8AF, UK
callum.stewart@kcl.ac.uk
Richard JB Dobson
Institute of Psychiatry,
Psychology & Neuroscience
King’s College London
London, SE5 8AF, UK
richard.j.dobson@kcl.ac.uk
Sebastian Böttcher
Epilepsy Center, Medical Center
- University of Freiburg
79106 Freiburg, Germany
sebastian.boettcher@uniklinik-
freiburg.de
Amos A Folarin
Institute of Psychiatry,
Psychology & Neuroscience
King’s College London
London, SE5 8AF, UK
amos.folarin@kcl.ac.uk
Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full citation
on the first page. Copyrights for third-party components of this work must be honored.
For all other uses, contact the Owner/Author.
UbiComp/ISWC ’18 Adjunct, October 8-12, 2018, Singapore, Singapore
© 2018 Copyright is held by the owner/author(s).
ACM ISBN 978-1-4503-5966-5/18/10.
https://doi.org/10.1145/3267305.3267578
Abstract
The traditional hospital set-up is not appropriate for long-
term epilepsy seizure detection in naturalistic ambulatory
settings. To explore the feasibility of seizure detection in
such a setting, an in-hospital study was conducted to eval-
uate three wearable devices and a data collection platform
for ambulatory seizure detection. The platform collects and
processes data for study administrators, clinicians and data
scientists, who use it to create models to detect seizures.
For that purpose, all data collected from the wearable de-
vices is additionally synchronized with the hospital EEG and
video, with gold-standard seizure labels provided by trained
clinicians. Data collected by wearable devices shows po-
tential for seizure detection in out-of-hospital based and
ambulatory settings.
Author Keywords
mHealth; mobile app; seizures; epilepsy; brain disorders
ACM Classification Keywords
H.5.m [Human-centered computing (HCC)]: Ubiquitous and
mobile computing.
Introduction
The 22m Euro Remote Assessment of Disease and Re-
lapse - Central Nervous System (RADAR-CNS) Innovative
Medicines Initiative (IMI2) is a research programme aimed
at developing novel methods and infrastructure for measur-
ing major depressive disorder, epilepsy, and multiple sclero-
sis [1]. The goals of RADAR-CNS are achieved through the
RADAR-base platform [2]. RADAR-base aims to provide
a highly extensible platform that enables remote stream-
ing data collection, secure data transmission and scalable
solutions for data storage, management and access. This
paper focuses on the Epilepsy which uses in-hospital de-
ployments of the platform to evaluate the capability of unob-
trusive wearable devices for seizure detection.
Epilepsy is a common serious neurological condition, af-
fecting around 6 million people in Europe. A person may
experience a seizure at any place in any situation (both
night and day). Wearable devices have the potential to de-
tect seizures in daily living conditions allowing the study
of factors and precursors influencing seizure onset. The
downstream opportunity of real-time streaming capabilities
of the RADAR-base platform is especially advantageous
in epilepsy, as it may form the basis of a system to prevent
e.g. sudden unexpected death in epilepsy (SUDEP).
The first application of RADAR-base is currently ongoing
in a trial of approximately 200 patients across two sites.
Patients are recruited prior to undergoing routine inpatient
video-EEG monitoring as part of their conventional care.
Patients are typically recorded for 5-7 days. During this pe-
riod, additional devices are worn by the patient including
the Empatica E4, Biovotion Everion, and Faros 180. The
concurrently recorded video-EEG provides a gold standard
against which the capability of these devices to detect gen-
eralized tonic-clonic and focal seizures can be evaluated.
Figure 1: Data flow for the
epilepsy study, enabling A)
investigation of data quality from
the wearable devices and B)
comparison of clinical information
from new data streams to gold
standard methods (EEG and
video).
Related Work
Many studies measuring non-EEG sensor seizure detec-
tion performance have been conducted [4]. The major-
Figure 2: System Overview: Wearable devices are streaming data
to RADAR-base platform in parallel to video-EEG.
ity of work has used accelerometers, however increasing
availability of different sensors has lead to a greater focus
on multimodal seizure detection [4]. Most studies so far
have been conducted in clinical environments or confined
to a single room and data often manually transferred [3].
However, Vandecasteele et. al. [5] used a commercially
available cloud-based platform to store privacy-sensitive
data. Similarly, Velez et. al. [6] transferred data over Blue-
tooth to a paired tablet device, which then uploaded to a
database over WiFi. However, only potential seizure events
detected by the wearable device were transferred. By con-
trast, the RADAR-base platform, provides a step change in
the capability for continuous high-frequency recordings from
commonly worn devices to be uploaded and processed in
real-time, while guaranteeing full control over the data and
platform architecture.
Methods and Procedures
The RADAR-CNS epilepsy case study involves collecting
data to a locally deployed server hosting the RADAR-base
platform, which synchronously collects data in parallel to
the video-EEG set-up. The requirements fulfilled by the
data collection apparatus are: (i) integration of several dif-
ferent wearable device types for separate collateral data
collection; (ii) capability to stream the device data in real-
time, with little or no patient interaction; (iii) easy manage-
ment of the involved devices for patients and study staff;
(iv) synchronisation of the wearable sensor data with the
video-EEG to an accuracy of 1/10s.
The wearable devices will route their data through an An-
droid device, one per patient. From there it is sent to the
platform server, for processing and availability to study staff.
In turn, the Android devices and video-EEG machines reg-
ularly synchronise with a common time-server. A detailed
clinical set-up is shown in Figures 1 and 2. Figure 3 shows
a patient in a hospital ward wearing several devices, and
the tablet streaming the sensor data to the platform.
The sensor set was established based on the need for sev-
eral parameters of interest such as heart rate variability (via
PPG, EKG), acceleration and electrodermal activity. De-
vices capable of streaming some or all of these included
the Empatica E4 (wrist), Biovotion Everion (upper arm), and
Faros 180 (chest). An additional factor in the choice of de-
vices was the sensor placement, which can affect the qual-
ity of the data produced. For example, there is greater mo-
ment of inertia at the wrist as compared to the upper arm or
chest, so movements generated by spasm during seizures
produce a stronger signal. Conversely, photoplethysmog-
raphy is sensitive to light and motion artefacts, resulting in
significant noise in recordings at the wrist, especially during
motor seizures.
Figure 3: In-hospital epilepsy
patient wearing devices (green),
connected to a tablet (red).
In addition, the selected devices are all capable of connect-
ing to an Android device and streaming the collected raw
data in real-time via a bluetooth connection. Removing the
need for future ambulatory patients to manage data down-
loads in their home and provides clinicians with a real-time
overview of the patient’s condition and research study man-
agement.
The Platform provides structured raw data that can easily
be processed, e.g. to analyse the utility and effectiveness
of the devices in epilepsy studies. Furthermore, raw data
can be used offline to train new seizure detection models,
which can be incorporated into the platform’s real-time data
processing infrastructure to test them on new data as it is
being collected. Thus the platform also provides the possi-
bility for real-time seizure detection and alerting.
Results
The presented set-up has been successfully tested in an
ongoing clinical trial at the video-EEG monitoring units of
both the Clinical Neurophysiology Department at King’s
College Hospital London and the Epilepsy Center at the
University Hospital of Freiburg. As of June 2018, 125 pa-
tients have been enrolled at both of the sites.
Where possible, participants have concurrently worn all
three devices, the Empatica, Biovotion and Faros. Figure 4
shows a three-minute segment of acceleration, blood vol-
ume pulse and electrodermal activity during a focal seizure
event. The motor component can clearly be seen in the
accelerometry, as well as the significant impact of the pa-
tients movements on the blood volume pulse signal, which
is generated by the photoplethysmography sensor of the
Empatica E4. Characteristic changes in all signals have
been observed during seizure events of several different
types (generalized tonic-clonic and focal seizures).
Figure 4: Empatica E4 sensor data. The shaded area indicates a
focal seizure with a motor component. (a) Accelerometer. (b)
Photoplethysmogram. (c) Electrodermal activity.
Conclusion and Future Work
We are investigating the potential of wearable devices as
clinically valuable alternatives to complement hospital-
based technologies, and as a prerequisite to future am-
bulatory passive remote monitoring of patients in their home
environment. The capabilities of the RADAR-base plat-
form are sufficient for an in-hospital study of patients with
epileptic seizures, and a further out-of-hospital ambulatory
is under planning.
Acknowledgements
This work has received support from the EU/EFPIA Inno-
vative Medicines Initiative Joint Undertaking 2 (RADAR-
CNS grant No 115902). We would like to acknowledge
The Hyve and RADAR-CNS Consortium (http://www.radar-
cns.org/partners) for their support. The Authors receive
funding support from the National Institute for Health Re-
search (NIHR) Biomedical Research Centre at South Lon-
don and Maudsley NHS Foundation Trust and KCL.
References
[1] 2017. RADAR-CNS. https://www.radar-cns.org/. (2017).
[2] 2018. RADAR-base. https://radar-base.org/. (2018).
[3] Judith Andel, Constantin Ungureanu, Johan Arends,
Francis Tan, Johannes Van Dijk, George Petkov,
Stiliyan Kalitzin, Thea Gutter, Al Weerd, Ben Vled-
der, Roland Thijs, Ghislaine Thiel, Kit Roes, and Frans
Leijten. 2017. Multimodal, automated detection of noc-
turnal motor seizures at home: Is a reliable seizure
detector feasible. Epilepsia Open (2017).
[4] Anouk Van de Vel, Kris Cuppens, Bert Bonroy, Milica
Milosevic, Katrien Jansen, Sabine Van Huffel, Bart
Vanrumste, Patrick Cras, Lieven Lagae, and Berten
Ceulemans. 2016. Non-EEG seizure detection sys-
tems and potential SUDEP prevention: State of the art:
Review and update. Seizure (2016).
[5] Kaat Vandecasteele, Thomas De Cooman, Ying Gu,
Evy Cleeren, Kasper Claes, Wim Van Paesschen,
Sabine Van Huffel, and Borbala Hunyadi. 2017. Auto-
mated Epileptic Seizure Detection Based on Wearable
ECG and PPG in a Hospital Environment. Sensors 17,
10 (2017).
[6] Mariel Velez, Robert S. Fisher, Victoria Bartlett, and
Scheherazade Le. 2016. Tracking generalized tonic-
clonic seizures with a wrist accelerometer linked to an
online database. Seizure (2016).
... Multiple instances of RADAR-base are deployed and in use for real-world studies of Epilepsy, Multiple Sclerosis and Major Depressive Disorder (MDD) under the umbrella of RADAR-CNS [20,21]. ...
... Latest participants have the facility to wear three devices (Faros, Biovotion, E4) concurrently. We have explained the detailed deployment of the platform for Epilepsy studies and initial collected data in our work [20]. ...
Article
Full-text available
ABSTRACT Background: With a wide range of use cases in both research and clinical domains, collecting continuous mobile health (mHealth) streaming data from multiple sources in a secure, highly scalable and extensible platform is of high interest to the open source mHealth community. The EU IMI RADAR-CNS program is an exemplar project with the requirements to support collection of high resolution data at scale; as such, the RADAR-base platform is designed to meet these needs and additionally facilitate a new generation of mHealth projects in this nascent field. Objective: Wide-bandwidth networks, smartphone penetrance and wearable sensors offer new possibilities for collecting (near) real-time high resolution datasets from large numbers of participants. We aimed to build a platform that would cater for large scale data collection for remote monitoring initiatives. Key criteria are around scalability, extensibility, security and privacy. Methods: RADAR-base is developed as a modular application, the backend is built on a backbone of the highly successful Confluent/Apache Kafka framework for streaming data. To facilitate scaling and ease of deployment, we use Docker containers to package the components of the platform. RADAR-base provides two main mobile apps for data collection, a Passive App and an Active App. Other 3rd Party Apps and sensors are easily integrated into the platform. Management user interfaces to support data collection and enrolment are also provided. Results: General principles of the platform components and design of RADAR-base are presented here, with examples of the types of data currently being collected from devices used in RADAR-CNS projects: Multiple Sclerosis, Epilepsy and Depression cohorts. Conclusions: RADAR-base is a fully functional, remote data collection platform built around Confluent/Apache Kafka and provides off-the-shelf components for projects interested in collecting mHealth datasets at scale.
... Multiple instances of RADAR-base are deployed and in use for real-world studies of Epilepsy, Multiple Sclerosis and Major Depressive Disorder (MDD) under the umbrella of RADAR-CNS [20,21]. ...
... Latest participants have the facility to wear three devices (Faros, Biovotion, E4) concurrently. We have explained the detailed deployment of the platform for Epilepsy studies and initial collected data in our work [20]. ...
Article
Full-text available
ABSTRACT Background: With a wide range of use cases in both research and clinical domains, collecting continuous mobile health (mHealth) streaming data from multiple sources in a secure, highly scalable and extensible platform is of high interest to the open source mHealth community. The EU IMI RADAR-CNS program is an exemplar project with the requirements to support collection of high resolution data at scale; as such, the RADAR-base platform is designed to meet these needs and additionally facilitate a new generation of mHealth projects in this nascent field. Objective: Wide-bandwidth networks, smartphone penetrance and wearable sensors offer new possibilities for collecting (near) real-time high resolution datasets from large numbers of participants. We aimed to build a platform that would cater for large scale data collection for remote monitoring initiatives. Key criteria are around scalability, extensibility, security and privacy. Methods: RADAR-base is developed as a modular application, the backend is built on a backbone of the highly successful Confluent/Apache Kafka framework for streaming data. To facilitate scaling and ease of deployment, we use Docker containers to package the components of the platform. RADAR-base provides two main mobile apps for data collection, a Passive App and an Active App. Other 3rd Party Apps and sensors are easily integrated into the platform. Management user interfaces to support data collection and enrolment are also provided. Results: General principles of the platform components and design of RADAR-base are presented here, with examples of the types of data currently being collected from devices used in RADAR-CNS projects: Multiple Sclerosis, Epilepsy and Depression cohorts. Conclusions: RADAR-base is a fully functional, remote data collection platform built around Confluent/Apache Kafka and provides off-the-shelf components for projects interested in collecting mHealth datasets at scale.
... We particularly faced the challenging issue of low compliance in the RADAR-CNS multi-disorder study [27] [25]. We now set about establishing reasons for the low compliance. ...
Conference Paper
Full-text available
The paper describes the various problems and challenges encountered during the development and remote data collection in a cross-platform hybrid application developed for remote monitoring of participants and what solutions were implemented to mitigate them. These problems and challenges are universal for hybrid applications and this paper digs deep into these in the domain of large-scale, long-duration mHealth research studies. From technical issues to issues with user compliance, this paper discusses the core problems inherent to these types of studies and technologies, and how to mitigate them.
Article
Full-text available
Electrocardiography has added value to automatically detect seizures in temporal lobe epilepsy (TLE) patients. The wired hospital system is not suited for a long-term seizure detection system at home. To address this need, the performance of two wearable devices, based on electrocardiography (ECG) and photoplethysmography (PPG), are compared with hospital ECG using an existing seizure detection algorithm. This algorithm classifies the seizures on the basis of heart rate features, extracted from the heart rate increase. The algorithm was applied to recordings of 11 patients in a hospital setting with 701 h capturing 47 (fronto-)temporal lobe seizures. The sensitivities of the hospital system, the wearable ECG device and the wearable PPG device were respectively 57%, 70% and 32%, with corresponding false alarms per hour of 1.92, 2.11 and 1.80. Whereas seizure detection performance using the wrist-worn PPG device was considerably lower, the performance using the wearable ECG is proven to be similar to that of the hospital ECG.
Article
Full-text available
Objective Automated seizure detection and alarming could improve quality of life and potentially prevent sudden, unexpected death in patients with severe epilepsy. As currently available systems focus on tonic–clonic seizures, we want to detect a broader range of seizure types, including tonic, hypermotor, and clusters of seizures. Methods In this multicenter, prospective cohort study, the nonelectroencephalographic (non‐EEG) signals heart rate and accelerometry were measured during the night in patients undergoing a diagnostic video‐EEG examination. Based on clinical video‐EEG data, seizures were classified and categorized as clinically urgent or not. Seizures included for analysis were tonic, tonic–clonic, hypermotor, and clusters of short myoclonic/tonic seizures. Features reflecting physiological changes in heart rate and movement were extracted. Detection algorithms were developed based on stepwise fulfillment of conditions during increases in either feature. A training set was used for development of algorithms, and an independent test set was used for assessing performance. Results Ninety‐five patients were included, but due to sensor failures, data from only 43 (of whom 23 patients had 86 seizures, representing 402 h of data) could be used for analysis. The algorithms yield acceptable sensitivities, especially for clinically urgent seizures (sensitivity = 71–87%), but produce high false alarm rates (2.3–5.7 per night, positive predictive value = 25–43%). There was a large variation in the number of false alarms per patient. Significance It seems feasible to develop a detector with high sensitivity, but false alarm rates are too high for use in clinical practice. For further optimization, personalization of algorithms may be necessary.
Article
Purpose Detection of, and alarming for epileptic seizures is increasingly demanded and researched. Our previous review article provided an overview of non-invasive, non-EEG (electro-encephalography) body signals that can be measured, along with corresponding methods, state of the art research, and commercially available systems. Three years later, many more studies and devices have emerged. Moreover, the boom of smart phones and tablets created a new market for seizure detection applications. Method We performed a thorough literature review and had contact with manufacturers of commercially available devices. Results This review article gives an updated overview of body signals and methods for seizure detection, international research and (commercially) available systems and applications. Reported results of non-EEG based detection devices vary between 2.2% and 100% sensitivity and between 0 and 3.23 false detections per hour compared to the gold standard video-EEG, for seizures ranging from generalized to convulsive or non-convulsive focal seizures with or without loss of consciousness. It is particularly interesting to include monitoring of autonomic dysfunction, as this may be an important pathophysiological mechanism of SUDEP (sudden unexpected death in epilepsy), and of movement, as many seizures have a motor component. Conclusion Comparison of research results is difficult as studies focus on different seizure types, timing (night versus day) and patients (adult versus pediatric patients). Nevertheless, we are convinced that the most effective seizure detection systems are multimodal, combining for example detection methods for movement and heart rate, and that devices should especially take into account the user⿿s seizure types and personal preferences.
Article
Purpose: Clinical management of epilepsy and current epilepsy therapy trials rely on paper or electronic diaries often with inaccurate self-reported seizure frequency as the primary outcome. This is the first study addressing the feasibility of detecting and recording generalized tonic-clonic seizures (GTCS) through a biosensor linked to an online seizure database. Method: A prospective trial was conducted with video-EEG (vEEG) in an epilepsy monitoring unit. Patients wore a wristwatch accelerometer that detected shaking and transmitted events via Bluetooth® to a bedside electronic tablet and then via Wi-Fi to an online portal. The watch recorded the date, time, audio, duration, frequency and amplitude of events. Events logged by the watch and recorded in a bedside paper diary were measured against vEEG, the "gold standard." Results: Thirty patients were enrolled and 62 seizures were recorded on vEEG: 31 convulsive and 31 non-convulsive. Twelve patients had a total of 31 convulsive seizures, and of those, 10 patients had 13 GTCS. The watch captured 12/13 (92.3%) GTCS. Watch audio recordings were consistent with seizures in 11/12 (91.6%). Data were successfully transferred to the bedside tablet in 11/12 (91.6%), and to the online database in 10/12 (83.3%) GTCS. The watch recorded 81 false positives, of which 42/81 (51%) were cancelled by the patients. Patients and caregivers verbally reported 15/62 seizures (24.2% sensitivity) but no seizures were recorded on paper logs. Conclusion: Automatic detection and recording of GTCS to an online database is feasible and may be more informative than seizure logging in a paper diary.
Anouk Van de Vel Kris Cuppens Bert Bonroy Milica Milosevic Katrien Jansen Sabine Van Huffel Bart Vanrumste Patrick Cras Lieven Lagae and Berten Ceulemans
Non-EEG seizure detection systems and potential SUDEP prevention: State of the art: Review and update
  • Anouk Van De Vel
  • Kris Cuppens
  • Bert Bonroy
  • Milica Milosevic
  • Katrien Jansen
  • Sabine Van Huffel
  • Bart Vanrumste
  • Patrick Cras
  • de Vel Anouk Van